Intelligent pain assessment method and system based on flexible electronic skin

Abstract

The application relates to a smart pain evaluation method and system based on a flexible electronic skin, and relates to the technical field of medical treatment.The method comprises the following steps: collecting physiological data of a plurality of sensing nodes through a flexible electronic skin; calculating the distance between the sensing nodes and a pain center to generate a spatial weight matrix; extracting pain coupling features according to the spatial weight matrix and the physiological data; fusing pain coupling features and patient portrait features by adopting a PA-Conv neural network to extract a focused feature map; extracting multi-dimensional features of a time sequence segment of the physiological data, and calculating the credibility of the time sequence segment; taking the credibility as a modulation gate, processing the focused feature map by adopting a time sequence coupling gate neural network, and outputting coupling time sequence features; inputting the coupling time sequence features into a pre-trained prior rule model to predict a pain type and obtain a smart pain score.The method can identify the correlation between pain spatial distribution and physiological response, and improve the recognition ability for pains in multiple parts and hidden pains.

Description

Technical Field

[0001] This invention belongs to the field of medical technology, specifically relating to an intelligent pain assessment method and system based on flexible electronic skin. Background Technology

[0002] In existing smart wards, pain assessment and care suffer from numerous technical challenges, with algorithmic issues being the core bottleneck. These challenges fail to meet the pain-free care needs of patients after surgery, with chronic diseases, and those with advanced illnesses. Current pain assessment methods utilize general physiological signal features, which only independently extract, globally statistically analyze, and indiscriminately stitch together EDA, EMG, and temperature signals. This approach completely deviates from the clinical reality of bedridden patients in smart wards. General features do not use pain location coordinates, making it impossible to distinguish between high-value signals at the pain center and interference signals far from the lesion. General features only reflect the overall signal amplitude and cannot characterize the relationship between spatial pain distribution and physiological response, resulting in extremely poor ability to identify multi-site and hidden pain. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent pain assessment method and system based on flexible electronic skin to improve the aforementioned problems. To achieve the above objective, the technical solution adopted by this invention is as follows:

[0004] In a first aspect, this application provides an intelligent pain assessment method based on flexible electronic skin, comprising:

[0005] Physiological data from multiple sensing nodes are collected using flexible electronic skin;

[0006] The distance between the sensor node and the pain center is calculated based on the coordinate information of the sensor node, and a spatial weight matrix is ​​generated.

[0007] Pain coupling features are extracted based on spatial weight matrix and physiological data;

[0008] Patient profile features are constructed, and the pain coupling features and patient profile features are fused using a PA-Conv neural network to extract focused feature maps. The PA-Conv neural network performs weighted pooling based on a spatial weight matrix.

[0009] Extract multidimensional features from time-series segments of physiological data and calculate the reliability of the time-series segments;

[0010] Using credibility as a modulation gate, a temporally coupled gated neural network is used to process the focused feature map and output coupled temporal features.

[0011] Coupled temporal features are input into a pre-trained prior rule model to predict pain type, and an intelligent pain score is obtained by mapping through the Sigmoid function.

[0012] Secondly, this application also provides an intelligent pain assessment system based on flexible electronic skin, comprising:

[0013] The first module is used to collect physiological data from multiple sensing nodes through flexible electronic skin;

[0014] The second module is used to calculate the distance between the sensor node and the pain center based on the coordinate information of the sensor node, and generate a spatial weight matrix.

[0015] The third module is used to extract pain coupling features based on the spatial weight matrix and physiological data;

[0016] The fourth module is used to construct patient profile features. It uses a PA-Conv neural network to fuse pain coupling features and patient profile features to extract focused feature maps. The PA-Conv neural network performs weighted pooling based on a spatial weight matrix.

[0017] The fifth module is used to extract multidimensional features from time-series segments of physiological data and calculate the reliability of the time-series segments;

[0018] The sixth module is used to use credibility as a modulation gate, employs a temporally coupled gated neural network to process the focused feature map, and outputs coupled temporal features;

[0019] The seventh module is used to input coupled temporal features into a pre-trained prior rule model to predict the type of pain, and obtain an intelligent pain score through Sigmoid function mapping.

[0020] The beneficial effects of this invention are as follows:

[0021] This invention is specifically designed for pain assessment in smart wards. It employs a dual-driven architecture of data and rules, utilizing deep learning's fitting capabilities while injecting prior physiological rules about pain into the prior rule model. This enables interpretable, non-black-box, and accurate pain assessment. Pain's physiological response exhibits strict spatial specificity. This invention allows each feature to simultaneously carry physiological information, spatial information, and pain correlation, enabling the identification of the correlation between spatial pain distribution and physiological response, thus improving the ability to recognize multi-site and hidden pain.

[0022] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart illustrating the intelligent pain assessment method based on flexible electronic skin, as described in this application.

[0025] Figure 2 This is a structural diagram of an intelligent pain assessment device based on flexible electronic skin, as described in an embodiment of this application.

[0026] Symbol explanation: 800 - Intelligent pain assessment device based on flexible electronic skin; 801 - Processor; 802 - Memory; 803 - Multimedia component; 804 - I / O interface; 805 - Communication component. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0028] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0029] Example 1:

[0030] See Figure 1 This application provides an intelligent pain assessment method based on flexible electronic skin, including steps S100, S200, S300, S400, S500, S600 and S700.

[0031] S100: Collects physiological data from multiple sensing nodes through flexible electronic skin;

[0032] The flexible electronic skin comprises 32 three-dimensional dot-matrix sensing nodes, employing a medical-grade stainless steel micro-spring suspension design (0.5mm telescopic range). It can accurately acquire raw signals required by the algorithm, including pain location information, 6-dimensional features of electrical skin analysis (EDA), 5-dimensional features of electromyography (EMG), and 3-dimensional features of local temperature. This provides high-fidelity data for algorithm coupling feature engineering, unlike existing sensors that only collect basic signals. The sensor incorporates a micro-signal conditioning chip, which can perform algorithm-customized db4 wavelet basis denoising and 20-500Hz bandpass filtering pre-processing to eliminate electromagnetic interference from the ward and non-painful muscle contraction noise in advance, reducing the computational load of the terminal algorithm and improving the algorithm inference speed.

[0033] S200. Calculate the distance between the sensor node and the pain center based on the coordinate information of the sensor node, and generate a spatial weight matrix;

[0034] The pain center is the center of the patient's lesion. Based on the coordinates of 32 lattice nodes, the Euclidean distance D between the sensing node and the pain center is calculated. i A pain spatial weight matrix W(x,y) is generated, where the signal weight increases with distance from the pain center. First, a distance threshold is set; when D... i When the distance is less than the threshold, its weight is directly set to 1; when D i When the distance is greater than or equal to the distance threshold, the weight is calculated according to the preset decay function.

[0035] S300. Extract pain coupling features based on the spatial weight matrix and physiological data, as follows:

[0036] S310. Weight the peak value of skin conductance according to the spatial weight matrix and calculate the weighted average value of conductance recovery time to obtain the EDA coupling characteristics.

[0037] S320. The root mean square of electromyography (EMG) is multiplied by a preset site contraction sensitivity, and the number of EMG zero-crossing points is weighted according to the spatial weight matrix to obtain the EMG coupling characteristics.

[0038] S330. The skin temperature change rate is weighted according to the spatial weight matrix, and the deviation between the measured temperature value and the steady-state reference value is calculated to obtain the temperature coupling characteristics.

[0039] S340. Calculate the amplitude crossover ratio and mutual information coefficient between skin conductance data, electromyography data and temperature data to obtain cross-coupling characteristics, including: (EDA rise time × EMG zero crossing number) / temperature change rate, EDA-EMG amplitude crossover ratio, and EMG-temperature mutual information coefficient.

[0040] S400. Construct patient profile features, and use PA-Conv (Position Adaptive Convolution) neural network to fuse pain coupling features and patient profile features to extract focused feature maps. The PA-Conv neural network performs weighted pooling based on the spatial weight matrix.

[0041] S410. Construct patient profile features based on the patient's age, cognitive ability, symptoms, and disease severity coefficient;

[0042] The patient's age is mapped to an age coefficient A∈[0,1], and the patient's cognitive ability is mapped to a cognitive coefficient Cog∈[0,1]; the symptoms are coded and the severity coefficient S∈[0,1] is determined;

[0043] S420. Construct a coupling feature map based on the pain coupling features of all sensor nodes; each sensor node corresponds to a pixel on the coupling feature map, and match the pain coupling features to the corresponding pixels to obtain the coupling feature map.

[0044] S430. Use a spatial weight matrix to perform weighted pooling on each position of the coupled feature map to obtain pain-weighted features;

[0045] That is, traverse all spatial locations of the coupled feature map, multiply the feature of each location by the pain weight of that location, sum all the results, divide by the sum of the weights to normalize, and the resulting vector is the pain-weighted feature.

[0046] S440. Perform a linear transformation on the pain-weighted features and patient profile features to generate gated channel weights;

[0047] The pain-weighted features and patient profile features are linearly transformed using the ReLU activation function, and then the transformed values ​​are reduced to between 0 and 1 using the Sigmoid function to form the gated channel weights.

[0048] S450. The original coupled feature map is weighted by gated channel weights to obtain a focused feature map.

[0049] S500: Extract multidimensional features from time-series segments of physiological data and calculate the reliability of the time-series segments;

[0050] Multidimensional features are obtained by calculating the slope of signal changes at each moment in the time series segment, the deviation from historical trends, and the consistency between adjacent peaks and valleys based on physiological data.

[0051] Multidimensional features are input into a predefined rule definition network, which outputs a credibility score. The rule definition network uses human prior knowledge (physiological rules) as soft constraints and embeds them into a learnable neural network structure. This network can use a constrained linear classifier, including a feature extraction layer, a rule encoding layer, and an output layer. In the rule encoding layer, human rules are transformed into constraints on the weights, so that they have both rule interpretability and can be fine-tuned through backpropagation of data. Finally, the credibility score in the interval [0,1] is output through the Sigmoid function.

[0052] The rule definition network needs to be pre-learned, but it is constrained by physiological intervals. For example, the confidence score will be lowered if the slope of the signal change changes drastically or the signal is out of sync with the known intervention rhythm.

[0053] Based on the credibility scores at each time point, credibility curves are plotted for the time series segments. These credibility curves clearly indicate which time periods contribute significantly to the final score (credible) and which are judged as interference (unreliable), thus achieving a non-black-box approach in the time series dimension.

[0054] S600: Using credibility as a modulation gate, a time-coupled gated neural network is used to process the focused feature map and output coupled time-series features;

[0055] The confidence level is used as a modulation gate to recalibrate the hidden state ht of the original time-coupled gated (TC-GLU) output:

[0056] ht′=T×ht;

[0057] Where ht is the original hidden state of the time-coupled gating output, ht′ is the hidden state after recalibration, and T is the confidence level.

[0058] S700: Input the coupled temporal features into the pre-trained prior rule model to predict the pain type, and obtain the intelligent pain score through Sigmoid function mapping.

[0059] The prior rule model maps extracted temporal features (slope of signal change, deviation from historical trends, and consistency between adjacent peaks and valleys) to specific pain types using pre-defined clinical physiological rules. Mapping tables can be pre-built; for example, high slope, low deviation, and moderate consistency are mapped to somatic pain, while high / very high slope, high deviation, and low consistency are mapped to neuropathic pain.

[0060] By utilizing coupled temporal features, pain types (such as somatic pain / neuralgia / cancer pain) are predicted, and the type probability is output. The contribution ratios of EDA, EMG, and temperature signals in subsequent feature fusion can be adjusted according to the pain type.

[0061] If the condition is diagnosed as neuralgia, abnormal electromyography (EMG) is a key indicator. The weight of the EMG channels can be multiplied by a gain factor (e.g., 1.5) to highlight the spontaneous activity characteristics of abnormal EMG. If the condition is diagnosed as somatic pain, EDA and local temperature changes are more representative, and their weights can be increased accordingly.

[0062] Calculate the age-cognition coupling term, cognition-pain coupling term, and age-pain coupling term based on the patient's age, cognitive ability, and pain type;

[0063] Age-cognitive coupling term: AC = A × Cog λ ;

[0064] Cognitive-pain coupling term: CP = Cog × P

[0065] Age-pain coupling term: AP = A × P θ ;

[0066] A represents age coefficient, Cog represents cognitive coefficient, P represents pain type, and λ represents pain level. θ is the coupling index obtained from learning clinical data.

[0067] Calculate the coupling confidence level based on the age-cognition coupling term, the cognition-pain coupling term, and the age-pain coupling term;

[0068] Gb=sigmoid(ω1·AC+ω2·CP+ω3·AP+b)

[0069] Calculate age credibility and cognitive credibility based on age and cognitive ability, respectively;

[0070] Age reliability: Ga=0, A<0.4;

[0071] Cognitive credibility: Gc=0, Cog<0.2;

[0072] Subjective credibility is calculated based on coupling credibility, age credibility, and cognitive credibility.

[0073] G = Gb·Ga·Gc; and satisfies the constraint: C ∈ [0,1]

[0074] The subjective pain score and intelligent pain score provided by the patient are enhanced and integrated based on the subjective credibility to obtain the final pain score.

[0075] Calculate the deviation between subjective pain score and intelligent pain score: Δ=|S o -S s | / 10 (Normalized scoring bias); S o For intelligent pain scoring, S s Subjective pain rating;

[0076] The bias penalty is: Δp = exp(-Δ·τ); τ is the learned penalty coefficient.

[0077] Final pain score after correction:

[0078] S=S s ·C·Δp+S o ·(1-C); C represents the nonlinear coupling confidence level.

[0079] By designing an original credibility calculation structure with nonlinear coupling, gating constraints, and patient profile adaptation, age, cognitive level, and pain type are coupled and gated, rather than nonlinearly superimposed; at the same time, a nonlinear correction function with bias suppression is provided to achieve accurate correction of subjective scores.

[0080] To address the noise interference that is prone to occur in ward scenarios, noise needs to be specifically injected during model training to enhance robustness in ward scenarios. Specifically:

[0081] Construct a ward noise template vector based on the electromagnetic interference of ward equipment;

[0082] Electromagnetic interference in hospital wards originates from medical equipment such as monitors, infusion pumps, ventilators, and electrocardiographs, and exhibits fixed scene characteristics: based on interference collected from actual hospital wards, noise is generated that is location-dependent, spatially sparse, and operates in a fixed frequency band.

[0083] Obtain labeled physiological signal data, generate sparse node masks based on spatial weight matrix, and add noise template vectors to the labeled physiological signal data based on sparse node masks to obtain training samples;

[0084] Based on the spatial weight matrix, noise is injected only in non-pain center regions to protect the pain center signal.

[0085] Pain coupling features are extracted from training samples and a PA-Conv neural network is trained.

[0086] Example 2:

[0087] This embodiment provides an intelligent pain assessment system based on flexible electronic skin, including:

[0088] The first module is used to collect physiological data from multiple sensing nodes through flexible electronic skin;

[0089] The second module is used to calculate the distance between the sensor node and the pain center based on the coordinate information of the sensor node, and generate a spatial weight matrix.

[0090] The third module is used to extract pain coupling features based on the spatial weight matrix and physiological data;

[0091] The fourth module is used to construct patient profile features. It uses a PA-Conv neural network to fuse pain coupling features and patient profile features, and extracts focused feature maps. The PA-Conv neural network performs weighted pooling based on a spatial weight matrix.

[0092] The fifth module is used to extract multidimensional features from time-series segments of physiological data and calculate the reliability of the time-series segments;

[0093] The sixth module is used to use credibility as a modulation gate, employs a temporally coupled gated neural network to process the focused feature map, and outputs coupled temporal features;

[0094] The seventh module is used to input coupled temporal features into a pre-trained prior rule model to predict the type of pain, and obtain an intelligent pain score through Sigmoid function mapping.

[0095] As an optional implementation, the physiological data includes skin conductance data, electromyography data, and temperature data, and the third module includes:

[0096] The first unit is used to weight the peak value of skin conductance according to the spatial weight matrix and calculate the weighted average value of conductance recovery time to obtain EDA coupling characteristics;

[0097] The second unit is used to multiply the root mean square of electromyography by a preset site contraction sensitivity, and weight the number of zero-crossing points of electromyography according to the spatial weight matrix to obtain EMG coupling characteristics.

[0098] The third unit is used to weight the rate of change of skin temperature according to the spatial weight matrix and calculate the deviation between the measured temperature value and the steady-state reference value to obtain the temperature coupling characteristics.

[0099] The fourth unit is used to calculate the amplitude crossover ratio and mutual information coefficient between skin conductance data, electromyography data, and temperature data to obtain cross-coupling characteristics.

[0100] As an optional implementation, the fourth module includes:

[0101] The fifth unit is used to construct patient profile features based on the patient's age, cognitive ability, symptoms, and disease severity coefficient;

[0102] The sixth unit is used to construct a coupling feature map based on the pain coupling features of all sensing nodes;

[0103] The seventh unit is used to perform weighted pooling on each position of the coupled feature map using a spatial weight matrix to obtain pain-weighted features;

[0104] The eighth unit is used to perform a linear transformation on the pain-weighted features and patient profile features to generate gated channel weights;

[0105] The ninth unit is used to perform channel weighting on the original coupled feature map using gated channel weights to obtain a focused feature map.

[0106] As an optional implementation, the fifth module includes:

[0107] The tenth unit is used to calculate the slope of signal change at each moment in the time series segment, the deviation from historical trends, and the consistency of adjacent peaks and valleys based on physiological data, thereby obtaining multidimensional features;

[0108] The eleventh unit is used to input multidimensional features into a pre-defined rule network and output a confidence score.

[0109] Unit 12 is used to plot the credibility curve of the time series segment based on the credibility score at each time point.

[0110] Example 3:

[0111] Corresponding to the above method embodiments, this embodiment also provides an intelligent pain assessment device based on flexible electronic skin. The intelligent pain assessment device based on flexible electronic skin described below can be referred to in correspondence with the intelligent pain assessment method based on flexible electronic skin described above.

[0112] Figure 2 This is a block diagram illustrating an intelligent pain assessment device 800 based on flexible electronic skin, according to an exemplary embodiment. Figure 2As shown, the intelligent pain assessment device 800 based on flexible electronic skin includes a processor 801 and a memory 802. The intelligent pain assessment device 800 may also include one or more of the following: a multimedia component 803, an input / output (I / O) interface 804, and a communication component 805. The processor 801 controls the overall operation of the intelligent pain assessment device 800 to complete all or part of the steps in the aforementioned intelligent pain assessment method based on flexible electronic skin. The memory 802 stores various types of data to support the operation of the intelligent pain assessment device 800. This data may include, for example, commands for any application or method operating on the intelligent pain assessment device 800, and application-related data such as contact data, sent and received messages, images, audio, video, etc. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0113] Multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touchscreen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals.

[0114] The received audio signals can be further stored in memory 802 or transmitted via communication component 805. The audio component also includes at least one speaker for outputting audio signals. I / O interface 804 provides an interface between processor 801 and other interface modules, such as a keyboard, mouse, buttons, etc. These buttons can be virtual or physical. Communication component 805 is used for wired or wireless communication between the flexible electronic skin-based intelligent pain assessment device 800 and other devices. Wireless communication includes, for example, Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination thereof; therefore, the corresponding communication component 805 may include a Wi-Fi module, a Bluetooth module, or an NFC module.

[0115] Example 4:

[0116] Corresponding to the above embodiment of the intelligent pain assessment method based on flexible electronic skin, this embodiment also provides a readable storage medium. The readable storage medium described below can be referred to in conjunction with the intelligent pain assessment method based on flexible electronic skin described above.

[0117] A readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described embodiment of the intelligent pain assessment method based on flexible electronic skin.

[0118] Specifically, the readable storage medium can be a USB flash drive, external hard drive, read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk, or any other readable storage medium capable of storing program code.

[0119] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0120] 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 smart pain assessment method based on flexible electronic skin, characterized in that, include: Physiological data from multiple sensing nodes are collected using flexible electronic skin; The distance between the sensor node and the pain center is calculated based on the coordinate information of the sensor node, and a spatial weight matrix is ​​generated. Pain coupling features are extracted based on spatial weight matrix and physiological data; Patient profile features are constructed, and the pain coupling features and patient profile features are fused using a PA-Conv neural network to extract focused feature maps. The PA-Conv neural network performs weighted pooling based on a spatial weight matrix. Extract multidimensional features from time-series segments of physiological data and calculate the reliability of the time-series segments; Using credibility as a modulation gate, a temporally coupled gated neural network is used to process the focused feature map and output coupled temporal features. Coupled temporal features are input into a pre-trained prior rule model to predict pain type, and an intelligent pain score is obtained by mapping through the Sigmoid function.

2. The intelligent pain assessment method based on flexible electronic skin according to claim 1, characterized in that, The physiological data includes skin conductance data, electromyography data, and temperature data. The extraction of pain coupling features based on the spatial weight matrix and physiological data includes: The peak values ​​of skin conductance are weighted according to the spatial weight matrix, and the weighted average of conductance recovery time is calculated to obtain the EDA coupling characteristics; The EMG coupling characteristics are obtained by multiplying the root mean square of electromyography (EMG) by a preset site contraction sensitivity and weighting the number of EMG zero-crossing points according to the spatial weight matrix. The skin temperature change rate is weighted according to the spatial weight matrix, and the deviation between the measured temperature value and the steady-state reference value is calculated to obtain the temperature coupling characteristics. The amplitude crossover ratio and mutual information coefficient between skin conductance data, electromyography data, and temperature data are calculated to obtain the cross-coupling characteristics.

3. The intelligent pain assessment method based on flexible electronic skin according to claim 1, characterized in that, Patient profile features are constructed, and a PA-Conv neural network is used to fuse pain coupling features and patient profile features to extract focused feature maps, including: Patient profile features are constructed based on the patient's age, cognitive ability, symptoms, and disease severity coefficient; A coupling feature map is constructed based on the pain coupling features of all sensing nodes; A spatial weight matrix is ​​used to perform weighted pooling at each location of the coupled feature map to obtain pain-weighted features; A linear transformation is performed on the pain-weighted features and patient profile features to generate gated channel weights; The original coupled feature map is weighted by gating channel weights to obtain a focused feature map.

4. The intelligent pain assessment method based on flexible electronic skin according to claim 1, characterized in that, Extract multidimensional features from time-series segments of physiological data and calculate the reliability of these segments, including: Multidimensional features are obtained by calculating the slope of signal changes at each moment in the time series segment, the deviation from historical trends, and the consistency between adjacent peaks and valleys based on physiological data. Input multidimensional features into a predefined rule-defined network and output a credibility score; The credibility curves of the time series segments are plotted based on the credibility scores at each time point.

5. The intelligent pain assessment method based on flexible electronic skin according to claim 1, characterized in that, The method includes: Calculate the age-cognition coupling term, cognition-pain coupling term, and age-pain coupling term based on the patient's age, cognitive ability, and pain type; Calculate the coupling confidence level based on the age-cognition coupling term, the cognition-pain coupling term, and the age-pain coupling term; Calculate age credibility and cognitive credibility based on age and cognitive ability, respectively; Subjective credibility is calculated based on coupling credibility, age credibility, and cognitive credibility. The subjective pain score provided by the patient and the intelligent pain score are enhanced and integrated based on the subjective credibility to obtain the final pain score.

6. The intelligent pain assessment method based on flexible electronic skin according to claim 1, characterized in that, The method includes: Construct a ward noise template vector based on the electromagnetic interference of ward equipment; Obtain labeled physiological signal data, generate sparse node masks based on spatial weight matrix, and add noise template vectors to the labeled physiological signal data based on sparse node masks to obtain training samples; Pain coupling features are extracted from training samples and a PA-Conv neural network is trained.

7. A smart pain assessment system based on flexible electronic skin, characterized in that, include: The first module is used to collect physiological data from multiple sensing nodes through flexible electronic skin; The second module is used to calculate the distance between the sensor node and the pain center based on the coordinate information of the sensor node, and generate a spatial weight matrix. The third module is used to extract pain coupling features based on the spatial weight matrix and physiological data; The fourth module is used to construct patient profile features. It uses a PA-Conv neural network to fuse pain coupling features and patient profile features, and extracts focused feature maps. The PA-Conv neural network performs weighted pooling based on a spatial weight matrix. The fifth module is used to extract multidimensional features from time-series segments of physiological data and calculate the reliability of the time-series segments; The sixth module is used to use credibility as a modulation gate, employs a temporally coupled gated neural network to process the focused feature map, and outputs coupled temporal features; The seventh module is used to input coupled temporal features into a pre-trained prior rule model to predict the type of pain, and obtain an intelligent pain score through Sigmoid function mapping.

8. The intelligent pain assessment system based on flexible electronic skin according to claim 7, characterized in that, The physiological data includes skin conductance data, electromyographic data, and temperature data. The third module includes: The first unit is used to weight the peak value of skin conductance according to the spatial weight matrix and calculate the weighted average value of conductance recovery time to obtain EDA coupling characteristics; The second unit is used to multiply the root mean square of electromyography by a preset site contraction sensitivity, and weight the number of zero-crossing points of electromyography according to the spatial weight matrix to obtain EMG coupling characteristics. The third unit is used to weight the rate of change of skin temperature according to the spatial weight matrix and calculate the deviation between the measured temperature value and the steady-state reference value to obtain the temperature coupling characteristics. The fourth unit is used to calculate the amplitude crossover ratio and mutual information coefficient between skin conductance data, electromyography data, and temperature data to obtain cross-coupling characteristics.

9. The intelligent pain assessment system based on flexible electronic skin according to claim 7, characterized in that, The fourth module includes: The fifth unit is used to construct patient profile features based on the patient's age, cognitive ability, symptoms, and disease severity coefficient; The sixth unit is used to construct a coupling feature map based on the pain coupling features of all sensing nodes; The seventh unit is used to perform weighted pooling on each position of the coupled feature map using a spatial weight matrix to obtain pain-weighted features; The eighth unit is used to perform a linear transformation on the pain-weighted features and patient profile features to generate gated channel weights; The ninth unit is used to perform channel weighting on the original coupled feature map using gated channel weights to obtain a focused feature map.

10. The intelligent pain assessment system based on flexible electronic skin according to claim 7, characterized in that, The fifth module includes: The tenth unit is used to calculate the slope of signal change at each moment in the time series segment, the deviation from historical trends, and the consistency of adjacent peaks and valleys based on physiological data, thereby obtaining multidimensional features; The eleventh unit is used to input multidimensional features into a pre-defined rule network and output a confidence score. Unit 12 is used to plot the credibility curve of the time series segment based on the credibility score at each time point.

Images