Pain sensing patch method and system based on deep learning and multi-modal sensing

Abstract

The invention discloses a pain sensing patch method and system based on deep learning and multi-modal sensing. The method comprises the following steps: acquiring basic sensing data of multiple modals of a target patient in real time; on the basis of a data baseline corresponding to the current time point, calibrating the basic sensing data to obtain calibrated sensing data; according to the calibration sensing data and a deep learning model, predicting a pain part and a pain degree corresponding to the target patient; determining a patch strategy corresponding to the target patient based on a planning algorithm according to the pain part and the pain degree; the patch applying strategy is used for limiting the patch applying position, the patch applying number and the medicine applying parameters of the pain medicine applying patch equipment on the target patient. Therefore, accurate pain management based on multi-modal data and deep learning can be realized, the pain relieving effect of the drug delivery patch equipment and the comfort level of a patient are improved, and the pain misjudgment and drug delivery risk are reduced.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a pain perception patch method and system based on deep learning and multimodal sensing. Background Art

[0002] With the rapid growth of demand for precision medicine and pain management, medical institutions are paying more and more attention to optimizing the pain relief effect and patient comfort of drug delivery patches through intelligent equipment. Existing technologies usually collect basic sensor data from patients, use simple threshold analysis or manual evaluation methods to judge the pain state, and determine the patch position and parameters based on fixed drug delivery rules to support pain relief. Existing solutions lack real-time calibration of multimodal sensor data and dynamic prediction of deep learning models, making it difficult to accurately identify the location and degree of pain and optimize patch strategies. Commonly used static drug delivery plans cannot adapt to complex pain scenarios, resulting in misjudgment of pain or improper drug delivery, limiting the pain relief effect and patient comfort. It can be seen that the existing technology has defects that need to be addressed urgently. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a pain perception patch method and system based on deep learning and multimodal sensing, which can realize precise pain management based on multimodal data and deep learning, improve the pain relief effect of drug delivery patch equipment and patient comfort, and reduce the risk of misjudgment of pain and drug delivery.

[0004] In order to solve the above technical problems, the first aspect of the present invention discloses a pain perception patch method based on deep learning and multimodal sensing, the method comprising:

[0005] Acquire multiple modal basic sensor data of target patients in real time;

[0006] Calibrate the basic sensor data based on the data baseline corresponding to the current time point to obtain calibrated sensor data;

[0007] Predicting the pain location and pain severity of the target patient based on the calibrated sensor data and the deep learning model;

[0008] According to the pain location and pain severity, based on the planning algorithm, the patch strategy corresponding to the target patient is determined; the patch strategy is used to limit the patch position, number of patches and drug delivery parameters of the pain drug delivery patch device on the target patient.

[0009] As an optional embodiment, in the first aspect of the present invention, the basic sensing data includes blood flow optical sensing data, bioimpedance sensing data, myoelectric sensing data and mechanical sensing data.

[0010] As an optional embodiment, in the first aspect of the present invention, the pain medication patch device includes an instruction transmission module and a medication module; the medication module includes an infusion pump.

[0011] As an optional embodiment, in the first aspect of the present invention, calibrating the basic sensor data based on the data baseline corresponding to the current time point to obtain the calibrated sensor data includes:

[0012] Determining at least one abnormal data in all the basic sensor data based on a data anomaly screening rule;

[0013] Determining a data calibration parameter corresponding to each of the basic sensor data according to the data value abnormality parameter corresponding to the abnormal data and the data baseline corresponding to the current time point;

[0014] The sum of the data calibration parameter and the basic data value corresponding to each basic sensor data is calculated to obtain the corresponding calibration sensor data.

[0015] As an optional embodiment, in the first aspect of the present invention, the determining of at least one abnormal data in all the basic sensor data based on the data abnormality screening rule includes:

[0016] Clustering all the basic sensor data based on acquisition time points to obtain multiple data groups; the time difference between the acquisition time points of any two basic sensor data in each data group is less than a preset time difference threshold;

[0017] For each basic sensor data of each data group, calculating a data difference between a data value corresponding to the basic sensor data and a reference data value;

[0018] Calculating an average of time differences between an acquisition time point corresponding to the basic sensor data and acquisition time points of all other basic sensor data in the data group to obtain a time weight corresponding to the basic sensor data;

[0019] Calculating a ratio of the data difference corresponding to the basic sensor data to the time weight to obtain a data value anomaly parameter corresponding to the basic sensor data;

[0020] All the basic sensor data whose abnormal data value parameters are greater than a preset parameter threshold are screened out to obtain at least one abnormal data.

[0021] As an optional embodiment, in the first aspect of the present invention, determining the data calibration parameter corresponding to each of the basic sensor data based on the data value abnormality parameter corresponding to the abnormal data and the data baseline corresponding to the current time point includes:

[0022] Calculating the average of the differences between the data value abnormality parameters corresponding to all the abnormal data and the preset parameter threshold to obtain the corresponding baseline correction parameters;

[0023] For each basic sensor data, obtaining a data baseline corresponding to the data type of the basic sensor data corresponding to the current time point;

[0024] According to the baseline correction parameters and the data baseline, based on a baseline correction algorithm corresponding to the basic sensor data, a data calibration parameter corresponding to the basic sensor data is determined.

[0025] As an optional embodiment, in the first aspect of the present invention, predicting the pain location and pain severity corresponding to the target patient based on the calibrated sensor data and the deep learning model includes:

[0026] Dividing all the calibration sensing data into a plurality of sensing data groups corresponding to part regions; wherein the sensing position of each calibration sensing data in the sensing data group is within the part region;

[0027] Inputting each sensor data set into a trained deep learning model to obtain a predicted pain level corresponding to each sensor data set; the deep learning model is trained using a training dataset comprising a plurality of training sensor data and corresponding pain level annotations;

[0028] Re-clustering all the sensor data groups based on a dynamic programming algorithm to obtain at least one associated data group set; the regional distance between the part regions corresponding to any two of the sensor data groups in the associated data group set is less than a distance threshold; and the average value of the predicted pain levels corresponding to any two of the sensor data groups in the associated data group set is greater than a preset pain level threshold;

[0029] The area corresponding to the sensor data group in the associated data group set is determined as the pain area corresponding to the target patient, and the predicted pain level corresponding to each pain area is determined as the corresponding pain level.

[0030] As an optional embodiment, in the first aspect of the present invention, determining the patch strategy corresponding to the target patient based on the pain location and pain severity and a planning algorithm includes:

[0031] For each pain site, obtaining the number of times the target patient has had a patch applied to the pain site in history from a history database;

[0032] determining at least one associated site that is pain-associated with the pain site;

[0033] Calculating an average of the pain levels corresponding to all the associated parts to obtain the associated pain level corresponding to the pain part;

[0034] Calculating a weighted sum of the pain level corresponding to the pain site, the number of historical patch applications, and the associated pain level to obtain a planning priority parameter corresponding to the pain site;

[0035] Determining the objective function as minimizing the sum of the number of patches corresponding to all the pain sites with the average value of the drug administration parameters;

[0036] Identify constraints including:

[0037] The higher the planning priority parameter, the higher the priority of the patch at the pain site, the more patches there are, and the higher the drug administration parameter;

[0038] The sum of the number of patches corresponding to all the pain sites is no greater than a threshold number of patches corresponding to the target patient; the threshold number of patches is proportional to the degree of disease progression of the target patient;

[0039] Based on a dynamic programming algorithm, according to the objective function and the constraints, the patch strategy corresponding to the target patient is iteratively calculated.

[0040] A second aspect of an embodiment of the present invention discloses a pain perception patch system based on deep learning and multimodal sensing, the system comprising:

[0041] An acquisition module is used to acquire basic sensor data of multiple modalities of target patients in real time;

[0042] A calibration module, configured to calibrate the basic sensor data based on a data baseline corresponding to a current time point to obtain calibrated sensor data;

[0043] A prediction module, configured to predict the pain location and pain severity of the target patient based on the calibrated sensor data and a deep learning model;

[0044] A determination module is used to determine the patch strategy corresponding to the target patient based on the planning algorithm according to the pain location and pain severity; the patch strategy is used to limit the patch position, number of patches and drug delivery parameters of the pain drug delivery patch device on the target patient.

[0045] As an optional embodiment, in the second aspect of the present invention, the basic sensing data includes blood flow optical sensing data, bioimpedance sensing data, myoelectric sensing data and mechanical sensing data.

[0046] As an optional embodiment, in the second aspect of the present invention, the pain medication patch device includes an instruction transmission module and a medication module; the medication module includes an infusion pump.

[0047] As an optional embodiment, in the second aspect of the present invention, the specific manner in which the calibration module calibrates the basic sensor data based on the data baseline corresponding to the current time point to obtain the calibrated sensor data includes:

[0048] Determining at least one abnormal data in all the basic sensor data based on a data anomaly screening rule;

[0049] Determining a data calibration parameter corresponding to each of the basic sensor data according to the data value abnormality parameter corresponding to the abnormal data and the data baseline corresponding to the current time point;

[0050] The sum of the data calibration parameter and the basic data value corresponding to each basic sensor data is calculated to obtain the corresponding calibration sensor data.

[0051] As an optional embodiment, in the second aspect of the present invention, the calibration module determines, based on the data anomaly screening rule, a specific manner of determining at least one abnormal data in all the basic sensor data, including:

[0052] Clustering all the basic sensor data based on acquisition time points to obtain multiple data groups; the time difference between the acquisition time points of any two basic sensor data in each data group is less than a preset time difference threshold;

[0053] For each basic sensor data of each data group, calculating a data difference between a data value corresponding to the basic sensor data and a reference data value;

[0054] Calculating an average of time differences between an acquisition time point corresponding to the basic sensor data and acquisition time points of all other basic sensor data in the data group to obtain a time weight corresponding to the basic sensor data;

[0055] Calculating a ratio of the data difference corresponding to the basic sensor data to the time weight to obtain a data value anomaly parameter corresponding to the basic sensor data;

[0056] All the basic sensor data whose abnormal data value parameters are greater than a preset parameter threshold are screened out to obtain at least one abnormal data.

[0057] As an optional embodiment, in the second aspect of the present invention, the calibration module determines the specific manner in which the data calibration parameters corresponding to each of the basic sensor data are determined based on the data value abnormality parameters corresponding to the abnormal data and the data baseline corresponding to the current time point, including:

[0058] Calculating the average of the differences between the data value abnormality parameters corresponding to all the abnormal data and the preset parameter threshold to obtain the corresponding baseline correction parameters;

[0059] For each basic sensor data, obtaining a data baseline corresponding to the data type of the basic sensor data corresponding to the current time point;

[0060] According to the baseline correction parameters and the data baseline, based on a baseline correction algorithm corresponding to the basic sensor data, a data calibration parameter corresponding to the basic sensor data is determined.

[0061] As an optional embodiment, in the second aspect of the present invention, the prediction module predicts the specific manner in which the pain location and pain severity corresponding to the target patient are predicted based on the calibration sensor data and the deep learning model, including:

[0062] Dividing all the calibration sensing data into a plurality of sensing data groups corresponding to part regions; wherein the sensing position of each calibration sensing data in the sensing data group is within the part region;

[0063] Inputting each sensor data set into a trained deep learning model to obtain a predicted pain level corresponding to each sensor data set; the deep learning model is trained using a training dataset comprising a plurality of training sensor data and corresponding pain level annotations;

[0064] Re-clustering all the sensor data groups based on a dynamic programming algorithm to obtain at least one associated data group set; the regional distance between the part regions corresponding to any two of the sensor data groups in the associated data group set is less than a distance threshold; and the average value of the predicted pain levels corresponding to any two of the sensor data groups in the associated data group set is greater than a preset pain level threshold;

[0065] The area corresponding to the sensor data group in the associated data group set is determined as the pain area corresponding to the target patient, and the predicted pain level corresponding to each pain area is determined as the corresponding pain level.

[0066] As an optional embodiment, in the second aspect of the present invention, the determination module determines the specific manner of the patch strategy corresponding to the target patient based on the planning algorithm according to the pain location and pain severity, including:

[0067] For each pain site, obtaining the number of times the target patient has had a patch applied to the pain site in history from a history database;

[0068] determining at least one associated site that is pain-associated with the pain site;

[0069] Calculating an average of the pain levels corresponding to all the associated parts to obtain the associated pain level corresponding to the pain part;

[0070] Calculating a weighted sum of the pain level corresponding to the pain site, the number of historical patch applications, and the associated pain level to obtain a planning priority parameter corresponding to the pain site;

[0071] Determining the objective function as minimizing the sum of the number of patches corresponding to all the pain sites with the average value of the drug administration parameters;

[0072] Identify constraints including:

[0073] The higher the planning priority parameter, the higher the priority of the patch at the pain site, the more patches there are, and the higher the drug administration parameter;

[0074] The sum of the number of patches corresponding to all the pain sites is no greater than a threshold number of patches corresponding to the target patient; the threshold number of patches is proportional to the degree of disease progression of the target patient;

[0075] Based on a dynamic programming algorithm, according to the objective function and the constraints, the patch strategy corresponding to the target patient is iteratively calculated.

[0076] The third aspect of the present invention discloses another pain perception patch system based on deep learning and multimodal sensing, the system comprising:

[0077] a memory storing executable program code;

[0078] a processor coupled to the memory;

[0079] The processor calls the executable program code stored in the memory to execute part or all of the steps in the pain perception patch method based on deep learning and multimodal sensing disclosed in the first aspect of the present invention.

[0080] The fourth aspect of the present invention discloses a computer storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute some or all of the steps in the pain perception patch method based on deep learning and multimodal sensing disclosed in the first aspect of the present invention.

[0081] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0082] The present invention obtains multi-modal basic sensor data of the target patient in real time and obtains calibrated sensor data based on the baseline calibration of the data at the current time point. It uses a deep learning model to predict the location and degree of pain, and combines it with a planning algorithm to determine the patient's patch strategy. This can achieve precise pain management based on multimodal data and deep learning, improve the pain relief effect of the drug delivery patch device and the patient's comfort, and reduce the risk of misjudgment of pain and drug delivery. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0084] Figure 1 This is a flow chart of a pain perception patch method based on deep learning and multimodal sensing disclosed in an embodiment of the present invention.

[0085] Figure 2 This is a structural diagram of a pain perception patch system based on deep learning and multimodal sensing disclosed in an embodiment of the present invention.

[0086] Figure 3 This is a structural diagram of another pain perception patch system based on deep learning and multimodal sensing disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0087] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0088] The terms "first," "second," and so on, in the description and claims of the present invention and the accompanying drawings are used to distinguish between different objects, not to describe a specific order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product, or device comprising a series of steps or elements is not limited to the listed steps or elements but may optionally include steps or elements not listed, or may optionally include other steps or elements inherent to the process, method, product, or device.

[0089] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute a separate or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0090] The present invention discloses a pain-sensing patch method and system based on deep learning and multimodal sensing. By acquiring multimodal basic sensor data from a target patient in real time and calibrating the sensor data based on the current time point data baseline, the method uses a deep learning model to predict the location and severity of pain. This method, combined with a planning algorithm, determines the patient's patch strategy. This method enables precise pain management based on multimodal data and deep learning, improves the pain relief effect of the drug delivery patch device and patient comfort, and reduces the risk of misdiagnosis of pain and drug delivery. These are described in detail below.

[0091] Example 1

[0092] See also Figure 1 , Figure 1 This is a flow chart of a pain perception patch method based on deep learning and multimodal sensing disclosed in an embodiment of the present invention. Figure 1 The described pain sensing patch method based on deep learning and multimodal sensing can be applied to a data processing system / data processing device / data processing server (wherein the server includes a local processing server or a cloud processing server). Figure 1 As shown, the pain sensing patch method based on deep learning and multimodal sensing may include the following operations:

[0093] 101. Acquire multiple modal basic sensor data of target patients in real time.

[0094] Optionally, the basic sensor data may be heart rate data, blood pressure data, electromyographic signal data, temperature data or motion data, which is not limited in the present invention.

[0095] Optionally, the multiple modalities may include physiological signal modality, motion signal modality or environmental signal modality, which is not limited in the present invention.

[0096] Optionally, the real-time acquisition can be achieved based on flexible electronic devices, wearable sensors or embedded monitoring devices, which is not limited in the present invention.

[0097] 102. Based on the data baseline corresponding to the current time point, calibrate the basic sensor data to obtain calibrated sensor data.

[0098] Optionally, the data baseline may be a historical average value, a standard value range, or a dynamically adjusted value, which is not limited in the present invention.

[0099] Optionally, the calibration process may be based on linear correction, normalization processing, or a model prediction algorithm, which is not limited in the present invention.

[0100] 103. Based on the calibrated sensor data and deep learning model, the corresponding pain location and pain degree of the target patient are predicted.

[0101] 104. According to the pain location and pain severity, based on the planning algorithm, determine the patch strategy corresponding to the target patient.

[0102] Optionally, the pain site may be the head, chest, abdomen, limbs or joint area, which is not limited in the present invention.

[0103] Optionally, the pain degree may be a pain grade, a pain score or a pain probability value, which is not limited in the present invention.

[0104] Optionally, the patch strategy is used to define the patch location, number of patches, and drug delivery parameters of the pain delivery patch device on the target patient.

[0105] It can be seen that the above-mentioned embodiment of the invention obtains the target patient's multi-modal basic sensor data in real time and obtains calibrated sensor data based on the baseline calibration of the data at the current time point, uses a deep learning model to predict the location and degree of pain, and combines the planning algorithm to determine the patient's patch strategy, thereby realizing precise pain management based on multimodal data and deep learning, improving the pain relief effect of the drug delivery patch device and the patient's comfort, and reducing the risk of misjudgment of pain and drug delivery.

[0106] As an optional embodiment, in the above steps, the basic sensing data includes blood flow optical sensing data, bioimpedance sensing data, myoelectric sensing data and mechanical sensing data.

[0107] It can be seen that through the above optional embodiments, the content of the sensor data is limited to comprehensively characterize the patient's real-time vital signs, assist in achieving precise pain management based on multimodal data and deep learning, improve the pain relief effect of the drug delivery patch device and the patient's comfort, and reduce the risk of misjudgment of pain and drug delivery.

[0108] As an optional embodiment, in the above steps, the pain medication patch device includes an instruction transmission module and a medication module; the medication module includes an infusion pump.

[0109] It can be seen that through the above optional embodiments, the technical details of the pain medication patch device are defined, so that the medication patch device can more accurately and intelligently implement the patient's pain link based on the patch strategy, assist in achieving precise pain management based on multimodal data and deep learning, improve the pain relief effect of the medication patch device and the patient's comfort, and reduce the risk of pain misjudgment and medication.

[0110] As an optional embodiment, in the above step, calibrating the basic sensor data based on the data baseline corresponding to the current time point to obtain the calibrated sensor data includes:

[0111] Based on the data anomaly screening rule, determining at least one abnormal data in all basic sensor data;

[0112] Determine the data calibration parameters corresponding to each basic sensor data based on the data value abnormality parameters corresponding to the abnormal data and the data baseline corresponding to the current time point;

[0113] The sum of the data calibration parameter and the basic data value corresponding to each basic sensor data is calculated to obtain the corresponding calibrated sensor data.

[0114] Optionally, the data anomaly screening rule may be a threshold-based rule, a statistical anomaly detection rule, or a machine learning classification rule, which is not limited in the present invention.

[0115] Optionally, the abnormal data may be data beyond a normal range, mutation data, or noise data, which is not limited in the present invention.

[0116] Optionally, the data value anomaly parameter may be a deviation value, an anomaly score, or an anomaly probability, which is not limited in the present invention.

[0117] Optionally, the data calibration parameter may be a correction coefficient, an offset, or a weighting factor, which is not limited in the present invention.

[0118] It can be seen that through the above optional embodiments, abnormal data in the basic sensor data is identified based on data anomaly screening rules, the calibration parameters of each basic sensor data are determined in combination with the data value abnormality parameters of the abnormal data and the data baseline of the current time point, and the sum of the calibration parameters and the basic data value is calculated to obtain the calibrated sensor data, thereby realizing precise data processing based on anomaly analysis and baseline calibration, improving the accuracy of pain prediction for target patients and the effectiveness of patch strategies, and reducing the risk of pain relief caused by data errors.

[0119] As an optional embodiment, in the above step, determining at least one abnormal data in all basic sensor data based on the data abnormality screening rule includes:

[0120] Clustering all basic sensor data based on acquisition time points to obtain multiple data groups; optionally, the time difference between the acquisition time points of any two basic sensor data in each data group is less than a preset time difference threshold;

[0121] For each basic sensor data of each data group, calculating the data difference between the data value corresponding to the basic sensor data and the reference data value;

[0122] Calculating the average of the time differences between the acquisition time point corresponding to the basic sensor data and the acquisition time points of all other basic sensor data in the data group to obtain a time weight corresponding to the basic sensor data;

[0123] Calculate the ratio of the data difference corresponding to the basic sensor data to the time weight to obtain the data value abnormality parameter corresponding to the basic sensor data;

[0124] All basic sensor data having abnormal parameters of data values ​​greater than a preset parameter threshold are screened out to obtain at least one abnormal data.

[0125] Optionally, the clustering may be based on K-means clustering, hierarchical clustering or time series clustering algorithm, which is not limited in the present invention.

[0126] Optionally, the reference data value may be a historical average value, a standard value, or a model prediction value, which is not limited in the present invention.

[0127] Optionally, the data value abnormality parameter can be a normalized score, an abnormality degree value or an abnormality probability, which is not limited in the present invention.

[0128] Optionally, the ratio may be calculated by linear mapping, logarithmic transformation or normalization processing, which is not limited in the present invention.

[0129] It can be seen that through the above optional embodiments, the basic sensor data are clustered based on the acquisition time point to form multiple data groups, the difference between the data value of each basic sensor data in each data group and the reference data value is calculated, the average time difference between the data acquisition time point and other data time points in the group is combined as the time weight, the ratio of the data difference to the time weight is calculated as the data value abnormality parameter, and the abnormal data beyond the parameter threshold is screened out, thereby realizing accurate abnormal data identification based on time clustering and weighted difference analysis, improving the accuracy of the target patient's calibration sensor data and the reliability of pain prediction, and reducing the risk of data abnormality interference.

[0130] As an optional embodiment, in the above steps, determining the data calibration parameters corresponding to each basic sensor data according to the data value abnormality parameters corresponding to the abnormal data and the data baseline corresponding to the current time point includes:

[0131] Calculate the average value of the difference between the data value abnormality parameter corresponding to all abnormal data and the preset parameter threshold to obtain the corresponding baseline correction parameter;

[0132] For each basic sensor data, obtain a data baseline corresponding to the data type of the basic sensor data corresponding to the current time point;

[0133] According to the baseline correction parameters and the data baseline, based on the baseline correction algorithm corresponding to the basic sensor data, the data calibration parameters corresponding to the basic sensor data are determined.

[0134] Optionally, the data baseline may be obtained based on historical data, preset standard values, or real-time model predictions, which is not limited in the present invention.

[0135] Optionally, the data baseline may be dynamically updated to reflect changes in patient status or environment, which is not limited in the present invention.

[0136] Optionally, the baseline correction algorithm may be a linear correction algorithm, a regression correction algorithm, or a machine learning correction model, which is not limited in the present invention.

[0137] It can be seen that through the above optional embodiments, the baseline correction parameters are obtained by calculating the average value of the difference between the data value abnormal parameters of all abnormal data and the preset parameter threshold, and the data calibration parameters are determined based on the baseline correction algorithm in combination with the data baseline of the corresponding data type of each basic sensor data at the current time point, thereby realizing accurate calibration parameter calculation based on the abnormal parameters and data baseline, improving the accuracy of the target patient's calibration sensor data and the reliability of pain prediction, and reducing the risk of pain relief deviation caused by data anomalies.

[0138] As an optional embodiment, in the above steps, predicting the pain location and pain severity of the target patient based on the calibrated sensor data and the deep learning model includes:

[0139] Dividing all calibration sensing data into sensing data groups corresponding to a plurality of part regions; optionally, the sensing position of each calibration sensing data in the sensing data group is within the part region;

[0140] Inputting each sensor data set into a trained deep learning model to obtain a predicted pain level corresponding to each sensor data set; optionally, the deep learning model is trained using a training dataset comprising a plurality of training sensor data and corresponding pain level annotations;

[0141] Re-clustering all sensor data groups based on a dynamic programming algorithm to obtain at least one associated data group set; optionally, a region distance between region areas corresponding to any two sensor data groups in the associated data group set is less than a distance threshold; an average value of predicted pain levels corresponding to any two sensor data groups in the associated data group set is greater than a preset pain level threshold;

[0142] The part area corresponding to the sensor data group in the associated data group set is determined as the pain part corresponding to the target patient, and the predicted pain level corresponding to each pain part is determined as the corresponding pain level.

[0143] Optionally, the deep learning model can be a convolutional neural network, a recurrent neural network or an attention mechanism model, which is not limited in the present invention.

[0144] It can be seen that through the above optional embodiments, by dividing the calibrated sensor data into sensor data groups corresponding to multiple part areas according to the sensor position, inputting the trained deep learning model to predict the pain level of each group, and re-clustering the data groups based on the dynamic programming algorithm to form a collection of associated data groups with more concentrated parts and higher pain levels, the part area of ​​the data group in the collection is determined as the pain part of the target patient and the corresponding predicted pain level is used as the pain level, thereby realizing accurate pain location and level identification based on part division and dynamic clustering, improving the accuracy of patch strategy formulation and pain relief effect, and reducing the risk of misjudgment of pain.

[0145] As an optional embodiment, in the above steps, determining the patch strategy corresponding to the target patient based on the planning algorithm according to the pain location and pain severity includes:

[0146] For each pain site, the target patient's historical number of patch applications at the pain site is obtained from the historical database;

[0147] determining at least one associated site that is pain-associated with the pain site;

[0148] Calculate the average of the pain levels corresponding to all associated parts to obtain the associated pain level corresponding to the pain part;

[0149] Calculate the weighted sum of the pain level, the number of historical patch applications, and the associated pain level corresponding to the pain site to obtain the planning priority parameter corresponding to the pain site;

[0150] The objective function is determined to be the sum of the number of patches corresponding to all pain sites to minimize the average value of the drug administration parameters;

[0151] Identify constraints including:

[0152] Pain sites with higher planning priority parameters had higher patch priority, more patches, and higher drug delivery parameters;

[0153] The sum of the number of patches corresponding to all pain sites is no greater than a threshold number of patches corresponding to the target patient; optionally, the threshold number of patches is proportional to the degree of disease progression of the target patient;

[0154] Based on the dynamic programming algorithm, the patch strategy corresponding to the target patient is iteratively calculated according to the objective function and constraints.

[0155] Optionally, the dynamic programming algorithm may be a recursive optimization algorithm, a particle swarm algorithm, a genetic immune algorithm, a state compression algorithm, or a hierarchical optimization algorithm, which is not limited in the present invention.

[0156] It can be seen that through the above optional embodiments, by obtaining the historical number of patches for each painful part of the target patient in the historical database, the associated parts associated with the part with pain are determined and their average pain levels are calculated as the associated pain levels, and the weighted sum of the pain levels, the historical number of patches and the associated pain levels is calculated as the planning priority parameter. The objective function is set to minimize the sum of the number of patches and the average value of the medication parameters. The constraints include that parts with high priority parameters have higher priorities, more patches, higher medication parameters and the total number of patches does not exceed the threshold value proportional to the degree of disease worsening. The patch strategy is iteratively calculated based on the dynamic programming algorithm, thereby realizing the formulation of accurate patch plans based on pain association and priority optimization, improving the patient's pain relief effect and comfort, and reducing the risk of overdose or underdose.

[0157] In a specific embodiment, a patch drug delivery system is implemented using the patch control scheme disclosed in the present invention, and its structural design includes:

[0158] Flexible electronic patch assembly: A miniaturized biosensor array is integrated on a multi-layer flexible substrate, including: high-precision electromyoelectric sensors for detecting muscle tension; nano-fabric sweat sensors for detecting cortisol and lactic acid; millimeter-wave radar for micro-movement monitoring of breathing / body movement frequency.

[0159] Edge computing module: The embedded AI chip runs a lightweight pain prediction model in real time, which can realize comprehensive pain level prediction by integrating five physiological characteristics such as EEG / EMG / GSR.

[0160] Clinical decision support system: Dynamically generates pain heat maps and links with the electronic medical record system to trigger medication recommendation threshold alarms.

[0161] Based on the flexible electronic patch component, the patient's multi-modal sensor data is obtained, and the prediction model set in the edge computing module is used to implement the technical steps disclosed in the present invention, which can effectively assist doctors in analyzing the patient's pain relief medication.

[0162] Example 2

[0163] See also Figure 2 , Figure 2 This is a schematic diagram of the structure of a pain sensing patch system based on deep learning and multimodal sensing disclosed in an embodiment of the present invention. Figure 2 The described pain sensing patch system based on deep learning and multimodal sensing can be applied to a data processing system / data processing device / data processing server (wherein the server includes a local processing server or a cloud processing server). Figure 2 As shown, the pain perception patch system based on deep learning and multimodal sensing may include:

[0164] The acquisition module 201 is used to acquire basic sensor data of multiple modalities of the target patient in real time.

[0165] The calibration module 202 is configured to calibrate the basic sensor data based on the data baseline corresponding to the current time point to obtain calibrated sensor data.

[0166] The prediction module 203 is used to predict the pain location and pain level of the target patient based on the calibrated sensor data and the deep learning model.

[0167] The determination module 204 is used to determine the patch strategy corresponding to the target patient based on the pain location and pain severity and the planning algorithm.

[0168] Optionally, the patch strategy is used to define the patch location, number of patches, and drug delivery parameters of the pain delivery patch device on the target patient.

[0169] It can be seen that the above-mentioned embodiment of the invention obtains the target patient's multi-modal basic sensor data in real time and obtains calibrated sensor data based on the baseline calibration of the data at the current time point, uses a deep learning model to predict the location and degree of pain, and combines the planning algorithm to determine the patient's patch strategy, thereby realizing precise pain management based on multimodal data and deep learning, improving the pain relief effect of the drug delivery patch device and the patient's comfort, and reducing the risk of misjudgment of pain and drug delivery.

[0170] As an optional embodiment, the basic sensing data includes blood flow optical sensing data, bioimpedance sensing data, myoelectric sensing data and mechanical sensing data.

[0171] It can be seen that through the above optional embodiments, the content of the sensor data is limited to comprehensively characterize the patient's real-time vital signs, assist in achieving precise pain management based on multimodal data and deep learning, improve the pain relief effect of the drug delivery patch device and the patient's comfort, and reduce the risk of misjudgment of pain and drug delivery.

[0172] As an optional embodiment, the pain medication patch device includes an instruction transmission module and a medication module; the medication module includes an infusion pump.

[0173] It can be seen that through the above optional embodiments, the technical details of the pain medication patch device are defined, so that the medication patch device can more accurately and intelligently implement the patient's pain link based on the patch strategy, assist in achieving precise pain management based on multimodal data and deep learning, improve the pain relief effect of the medication patch device and the patient's comfort, and reduce the risk of pain misjudgment and medication.

[0174] As an optional embodiment, the calibration module calibrates the basic sensor data based on the data baseline corresponding to the current time point to obtain the calibrated sensor data in a specific manner including:

[0175] Based on the data anomaly screening rule, determining at least one abnormal data in all basic sensor data;

[0176] Determine the data calibration parameters corresponding to each basic sensor data based on the data value abnormality parameters corresponding to the abnormal data and the data baseline corresponding to the current time point;

[0177] The sum of the data calibration parameter and the basic data value corresponding to each basic sensor data is calculated to obtain the corresponding calibrated sensor data.

[0178] It can be seen that through the above optional embodiments, abnormal data in the basic sensor data is identified based on data anomaly screening rules, the calibration parameters of each basic sensor data are determined in combination with the data value abnormality parameters of the abnormal data and the data baseline of the current time point, and the sum of the calibration parameters and the basic data value is calculated to obtain the calibrated sensor data, thereby realizing precise data processing based on anomaly analysis and baseline calibration, improving the accuracy of pain prediction for target patients and the effectiveness of patch strategies, and reducing the risk of pain relief caused by data errors.

[0179] As an optional embodiment, the calibration module determines a specific method of detecting at least one abnormal data in all basic sensor data based on the data abnormality screening rule, including:

[0180] Clustering all basic sensor data based on acquisition time points to obtain multiple data groups; optionally, the time difference between the acquisition time points of any two basic sensor data in each data group is less than a preset time difference threshold;

[0181] For each basic sensor data of each data group, calculating the data difference between the data value corresponding to the basic sensor data and the reference data value;

[0182] Calculating the average of the time differences between the acquisition time point corresponding to the basic sensor data and the acquisition time points of all other basic sensor data in the data group to obtain a time weight corresponding to the basic sensor data;

[0183] Calculate the ratio of the data difference corresponding to the basic sensor data to the time weight to obtain the data value abnormality parameter corresponding to the basic sensor data;

[0184] All basic sensor data having abnormal parameters of data values ​​greater than a preset parameter threshold are screened out to obtain at least one abnormal data.

[0185] It can be seen that through the above optional embodiments, the basic sensor data are clustered based on the acquisition time point to form multiple data groups, the difference between the data value of each basic sensor data in each data group and the reference data value is calculated, the average time difference between the data acquisition time point and other data time points in the group is combined as the time weight, the ratio of the data difference to the time weight is calculated as the data value abnormality parameter, and the abnormal data beyond the parameter threshold is screened out, thereby realizing accurate abnormal data identification based on time clustering and weighted difference analysis, improving the accuracy of the target patient's calibration sensor data and the reliability of pain prediction, and reducing the risk of data abnormality interference.

[0186] As an optional embodiment, the calibration module determines the specific method of the data calibration parameter corresponding to each basic sensor data according to the data value abnormality parameter corresponding to the abnormal data and the data baseline corresponding to the current time point, including:

[0187] Calculate the average value of the difference between the data value abnormality parameter corresponding to all abnormal data and the preset parameter threshold to obtain the corresponding baseline correction parameter;

[0188] For each basic sensor data, obtain a data baseline corresponding to the data type of the basic sensor data corresponding to the current time point;

[0189] According to the baseline correction parameters and the data baseline, based on the baseline correction algorithm corresponding to the basic sensor data, the data calibration parameters corresponding to the basic sensor data are determined.

[0190] It can be seen that through the above optional embodiments, the baseline correction parameters are obtained by calculating the average value of the difference between the data value abnormal parameters of all abnormal data and the preset parameter threshold, and the data calibration parameters are determined based on the baseline correction algorithm in combination with the data baseline of the corresponding data type of each basic sensor data at the current time point, thereby realizing accurate calibration parameter calculation based on the abnormal parameters and data baseline, improving the accuracy of the target patient's calibration sensor data and the reliability of pain prediction, and reducing the risk of pain relief deviation caused by data anomalies.

[0191] As an optional embodiment, the prediction module predicts the specific manner in which the target patient's corresponding pain location and pain severity are predicted based on the calibrated sensor data and the deep learning model, including:

[0192] Dividing all calibration sensing data into sensing data groups corresponding to a plurality of part regions; optionally, the sensing position of each calibration sensing data in the sensing data group is within the part region;

[0193] Inputting each sensor data set into a trained deep learning model to obtain a predicted pain level corresponding to each sensor data set; optionally, the deep learning model is trained using a training dataset comprising a plurality of training sensor data and corresponding pain level annotations;

[0194] Re-clustering all sensor data groups based on a dynamic programming algorithm to obtain at least one associated data group set; optionally, a region distance between region areas corresponding to any two sensor data groups in the associated data group set is less than a distance threshold; an average value of predicted pain levels corresponding to any two sensor data groups in the associated data group set is greater than a preset pain level threshold;

[0195] The part area corresponding to the sensor data group in the associated data group set is determined as the pain part corresponding to the target patient, and the predicted pain level corresponding to each pain part is determined as the corresponding pain level.

[0196] It can be seen that through the above optional embodiments, by dividing the calibrated sensor data into sensor data groups corresponding to multiple part areas according to the sensor position, inputting the trained deep learning model to predict the pain level of each group, and re-clustering the data groups based on the dynamic programming algorithm to form a collection of associated data groups with more concentrated parts and higher pain levels, the part area of ​​the data group in the collection is determined as the pain part of the target patient and the corresponding predicted pain level is used as the pain level, thereby realizing accurate pain location and level identification based on part division and dynamic clustering, improving the accuracy of patch strategy formulation and pain relief effect, and reducing the risk of misjudgment of pain.

[0197] As an optional embodiment, the determination module determines the specific manner of the patch strategy corresponding to the target patient based on the pain location and pain severity and the planning algorithm, including:

[0198] For each pain site, the target patient's historical number of patch applications at the pain site is obtained from the historical database;

[0199] determining at least one associated site that is pain-associated with the pain site;

[0200] Calculate the average of the pain levels corresponding to all associated parts to obtain the associated pain level corresponding to the pain part;

[0201] Calculate the weighted sum of the pain level, the number of historical patch applications, and the associated pain level corresponding to the pain site to obtain the planning priority parameter corresponding to the pain site;

[0202] The objective function is determined to be the sum of the number of patches corresponding to all pain sites to minimize the average value of the drug administration parameters;

[0203] Identify constraints including:

[0204] Pain sites with higher planning priority parameters had higher patch priority, more patches, and higher drug delivery parameters;

[0205] The sum of the number of patches corresponding to all pain sites is no greater than a threshold number of patches corresponding to the target patient; optionally, the threshold number of patches is proportional to the degree of disease progression of the target patient;

[0206] Based on the dynamic programming algorithm, the patch strategy corresponding to the target patient is iteratively calculated according to the objective function and constraints.

[0207] It can be seen that through the above optional embodiments, by obtaining the historical number of patches for each painful part of the target patient in the historical database, the associated parts associated with the part with pain are determined and their average pain levels are calculated as the associated pain levels, and the weighted sum of the pain levels, the historical number of patches and the associated pain levels is calculated as the planning priority parameter. The objective function is set to minimize the sum of the number of patches and the average value of the medication parameters. The constraints include that parts with high priority parameters have higher priorities, more patches, higher medication parameters and the total number of patches does not exceed the threshold value proportional to the degree of disease worsening. The patch strategy is iteratively calculated based on the dynamic programming algorithm, thereby realizing the formulation of accurate patch plans based on pain association and priority optimization, improving the patient's pain relief effect and comfort, and reducing the risk of overdose or underdose.

[0208] Example 3

[0209] See also Figure 3 , Figure 3 This is another pain perception patch system based on deep learning and multimodal sensing disclosed in an embodiment of the present invention. Figure 3The described pain sensing patch system based on deep learning and multimodal sensing is applied to a data processing system / data processing device / data processing server (wherein the server includes a local processing server or a cloud processing server). Figure 3 As shown, the pain perception patch system based on deep learning and multimodal sensing may include:

[0210] A memory 301 storing executable program code;

[0211] a processor 302 coupled to the memory 301;

[0212] The processor 302 calls the executable program code stored in the memory 301 to execute the steps of the pain perception patch method based on deep learning and multimodal sensing described in the first embodiment.

[0213] Example 4

[0214] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program enables a computer to execute the steps of the pain perception patch method based on deep learning and multimodal sensing described in Example 1.

[0215] Example 5

[0216] An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute the steps of the pain perception patch method based on deep learning and multimodal sensing described in Example 1.

[0217] The foregoing description of specific embodiments of the present disclosure is intended to illustrate a method for performing a multi-tasking process. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0218] The systems, devices, modules, or units described in the above embodiments may be implemented by computer chips or entities, or by products having certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0219] For the convenience of description, the above devices are described as being divided into various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0220] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the embodiments of this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0221] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0222] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0223] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0224] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0225] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0226] Computer-readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0227] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0228] This specification may be described in the general context of computer-executable instructions, such as program modules, executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media, including storage devices.

[0229] The various embodiments in this specification are described in a progressive manner. Similar parts between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the system embodiments are generally similar to the method embodiments, so the description is relatively simple. For relevant parts, refer to the description of the method embodiments.

[0230] Finally, it should be noted that the pain perception patch method and system based on deep learning and multimodal sensing disclosed in the embodiments of the present invention are only preferred embodiments of the present invention, and are only used to illustrate the technical solutions of the present invention, rather than to limit them. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A pain sensing patch method based on deep learning and multimodal sensing, characterized in that: The method comprises: Acquire multiple modal basic sensor data of target patients in real time; Calibrate the basic sensor data based on the data baseline corresponding to the current time point to obtain calibrated sensor data; Predicting the pain location and pain severity of the target patient based on the calibrated sensor data and the deep learning model; According to the pain location and pain severity, based on the planning algorithm, the patch strategy corresponding to the target patient is determined; the patch strategy is used to limit the patch position, number of patches and drug delivery parameters of the pain drug delivery patch device on the target patient.

2. The pain perception patch method based on deep learning and multimodal sensing according to claim 1, characterized in that: The basic sensing data includes blood flow optical sensing data, bioimpedance sensing data, myoelectric sensing data and mechanical sensing data.

3. The pain perception patch method based on deep learning and multimodal sensing according to claim 1, characterized in that: The pain medication patch device includes an instruction transmission module and a medication module; the medication module includes an infusion pump.

4. The pain perception patch method based on deep learning and multimodal sensing according to claim 1, characterized in that: The step of calibrating the basic sensor data based on the data baseline corresponding to the current time point to obtain the calibrated sensor data includes: Determining at least one abnormal data in all the basic sensor data based on a data anomaly screening rule; Determining a data calibration parameter corresponding to each of the basic sensor data according to the data value abnormality parameter corresponding to the abnormal data and the data baseline corresponding to the current time point; The sum of the data calibration parameter and the basic data value corresponding to each basic sensor data is calculated to obtain the corresponding calibration sensor data.

5. The pain perception patch method based on deep learning and multimodal sensing according to claim 4, characterized in that: The step of determining at least one abnormal data in all the basic sensor data based on the data abnormality screening rule includes: Clustering all the basic sensor data based on acquisition time points to obtain multiple data groups; the time difference between the acquisition time points of any two basic sensor data in each data group is less than a preset time difference threshold; For each basic sensor data of each data group, calculating a data difference between a data value corresponding to the basic sensor data and a reference data value; Calculating an average of time differences between an acquisition time point corresponding to the basic sensor data and acquisition time points of all other basic sensor data in the data group to obtain a time weight corresponding to the basic sensor data; Calculating a ratio of the data difference corresponding to the basic sensor data to the time weight to obtain a data value anomaly parameter corresponding to the basic sensor data; All the basic sensor data whose abnormal data value parameters are greater than a preset parameter threshold are screened out to obtain at least one abnormal data.

6. The pain perception patch method based on deep learning and multimodal sensing according to claim 4, characterized in that: The determining of the data calibration parameters corresponding to each of the basic sensing data according to the data value abnormality parameters corresponding to the abnormal data and the data baseline corresponding to the current time point includes: Calculating the average of the differences between the data value abnormality parameters corresponding to all the abnormal data and the preset parameter threshold to obtain the corresponding baseline correction parameters; For each basic sensor data, obtaining a data baseline corresponding to the data type of the basic sensor data corresponding to the current time point; According to the baseline correction parameters and the data baseline, based on a baseline correction algorithm corresponding to the basic sensor data, a data calibration parameter corresponding to the basic sensor data is determined.

7. The pain perception patch method based on deep learning and multimodal sensing according to claim 1, characterized in that: The step of predicting the pain location and pain severity of the target patient based on the calibrated sensor data and the deep learning model includes: Dividing all the calibration sensing data into a plurality of sensing data groups corresponding to part regions; wherein the sensing position of each calibration sensing data in the sensing data group is within the part region; Inputting each sensor data set into a trained deep learning model to obtain a predicted pain level corresponding to each sensor data set; the deep learning model is trained using a training dataset comprising a plurality of training sensor data and corresponding pain level annotations; Re-clustering all the sensor data groups based on a dynamic programming algorithm to obtain at least one associated data group set; the regional distance between the part regions corresponding to any two of the sensor data groups in the associated data group set is less than a distance threshold; and the average value of the predicted pain levels corresponding to any two of the sensor data groups in the associated data group set is greater than a preset pain level threshold; The area corresponding to the sensor data group in the associated data group set is determined as the pain area corresponding to the target patient, and the predicted pain level corresponding to each pain area is determined as the corresponding pain level.

8. The pain perception patch method based on deep learning and multimodal sensing according to claim 1, characterized in that: Determining the patch strategy corresponding to the target patient based on the pain location and pain severity and a planning algorithm includes: For each pain site, obtaining the number of times the target patient has had a patch applied to the pain site in history from a history database; determining at least one associated site that is pain-associated with the pain site; Calculating an average of the pain levels corresponding to all the associated parts to obtain the associated pain level corresponding to the pain part; Calculating a weighted sum of the pain level corresponding to the pain site, the number of historical patch applications, and the associated pain level to obtain a planning priority parameter corresponding to the pain site; Determining the objective function as minimizing the sum of the number of patches corresponding to all the pain sites with the average value of the drug administration parameters; Identify constraints including: The higher the planning priority parameter, the higher the priority of the patch at the pain site, the more patches there are, and the higher the drug administration parameter; The sum of the number of patches corresponding to all the pain sites is no greater than a threshold number of patches corresponding to the target patient; the threshold number of patches is proportional to the degree of disease progression of the target patient; Based on a dynamic programming algorithm, according to the objective function and the constraints, the patch strategy corresponding to the target patient is iteratively calculated.

9. A pain sensing patch system based on deep learning and multimodal sensing, characterized in that: The system comprises: An acquisition module is used to acquire basic sensor data of multiple modalities of target patients in real time; A calibration module, configured to calibrate the basic sensor data based on a data baseline corresponding to a current time point to obtain calibrated sensor data; A prediction module, configured to predict the pain location and pain severity of the target patient based on the calibrated sensor data and a deep learning model; A determination module is used to determine the patch strategy corresponding to the target patient based on the planning algorithm according to the pain location and pain severity; the patch strategy is used to limit the patch position, number of patches and drug delivery parameters of the pain drug delivery patch device on the target patient.

10. A pain sensing patch system based on deep learning and multimodal sensing, characterized in that: The system comprises: a memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the pain perception patch method based on deep learning and multimodal sensing as described in any one of claims 1 to 8.

Images