Ai-based personalized pain management system and method

Abstract

An AI-based personalized pain management system and method. The system comprises: a multi-dimensional data collection module, which is used for acquiring, in real time, multi-dimensional monitoring data including physiological state data, motion data, location information data, and bioelectric signal sensing data of a user; a pain type recognition module, which is used for performing feature extraction on the physiological state data and the bioelectric signal sensing data, and recognizing a pain type and a pain level of the user in view of basic information of the user; a pain trend prediction module, which is used for performing time series analysis on the basis of the pain type, the pain level and the multi-dimensional monitoring data, so as to predict a pain trend of the user in a future preset time period; and a feedback evaluation module, which is used for outputting a pain management evaluation report of the user on the basis of the pain type, the pain level and the pain trend. The system can accurately monitor, evaluate and intervene pain, so that pain management is more intelligent and personalized, thus helping to improve the quality of life of a user.

Description

A Personalized Pain Management System and Method Based on AI Technical Field

[0001] This invention relates to the field of health data processing technology, and in particular to an AI-based personalized pain management system and method. Background Technology

[0002] The dynamic changes in pain are often difficult to monitor and predict accurately, especially in patients with long-term chronic pain. Traditional methods rely on subjective patient feedback, which may be biased or inaccurate. With the development of sensor technology, wearable devices, and AI algorithms, it has become possible to collect bioelectrical signals and multidimensional physiological data in real time. In-depth analysis of this data can not only identify the type and level of pain but also predict future pain trends, thus providing a scientific basis for pain management.

[0003] Currently, some wearable devices are used to monitor users' physiological status and movement data, and some can also monitor pain status in real time. However, these devices have a low level of intelligence and lack in-depth personalized analysis and prediction functions. The dynamic changes and trends of pain are difficult to predict, especially for patients with chronic pain, whose pain may recur or suddenly worsen. Existing methods cannot effectively provide long-term, continuous monitoring and prediction. At the same time, each patient's physiological status, pain tolerance, and lifestyle habits are different, and traditional methods usually cannot provide personalized pain management plans based on individual differences.

[0004] Therefore, there is a need to propose an AI-based personalized pain management system and method that can achieve precise monitoring, assessment and intervention of pain, making pain management more intelligent and personalized, and helping to improve users' quality of life. Summary of the Invention

[0005] In view of this, the present invention provides an AI-based personalized pain management system and method to solve the technical problems of current pain assessment methods relying on subjective judgment, lacking multidimensional data support and real-time monitoring, resulting in a lack of personalized prediction and real-time feedback in the pain management process.

[0006] To achieve the above-mentioned technical objectives, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides an AI-based personalized pain management system, comprising:

[0008] The multidimensional data acquisition module is used to collect multidimensional monitoring data in real time, including user physiological state data, motion data, location information data, and bioelectrical signal sensing data.

[0009] The pain type recognition module is used to extract features from physiological state data and bioelectric signal sensing data, and combine them with the user's basic information to identify the user's pain type and pain level.

[0010] The pain trend prediction module is used to perform time series analysis based on pain type, pain level and multidimensional monitoring data to predict the user's pain trend within a preset time period in the future;

[0011] The feedback assessment module is used to output a pain management assessment report for users based on pain type, pain level, and pain trend.

[0012] Furthermore, the feature extraction from physiological state data and bioelectrical signal sensing data, combined with basic user information, to identify the user's pain type and pain level includes:

[0013] Extract SDNN and LF / HF power ratio from physiological state data;

[0014] Stress response values ​​were extracted from bioelectric signal sensing data.

[0015] The SDNN, LF / HF power ratio, and stress response values ​​are used as feature vectors and input into a trained KNN classifier to determine the pain type.

[0016] Based on the type of pain, a linear regression method is used to output a pain level score.

[0017] Furthermore, the process of performing time-series analysis based on pain type, pain level, and multidimensional monitoring data to predict the user's pain trend within a preset time period includes:

[0018] Multidimensional monitoring data is divided into time series data and static feature data;

[0019] Dynamic time warping is used to align time steps in time series data to construct a multidimensional feature vector.

[0020] Static feature data and multidimensional feature vectors are input into a prediction model based on an LSTM network for time series analysis to determine the pain level of users in each pain category within a preset future time period.

[0021] Furthermore, the structure of the prediction model based on the LSTM network is as follows:

[0022] An attention layer is added after the output of the base LSTM layer to dynamically assign weights to the output at each time step;

[0023] The static features and LSTM output are concatenated through a fully connected layer, and the concatenated data is mapped to the predicted pain score.

[0024] The model's loss function is the weighted mean squared error, and L2 regularization constraints are introduced.

[0025] Furthermore, the pain type identification module is deployed on a local server, the pain trend prediction module is deployed on a cloud server, and model compression is achieved through knowledge distillation.

[0026] Furthermore, the output user's pain management assessment report includes:

[0027] The prediction results are dynamically displayed using heatmaps and dynamic trend charts;

[0028] If the level of a preset pain type exceeds the risk threshold within a future time period, the user's multidimensional monitoring data will be uploaded to the medical institution.

[0029] Furthermore, the physiological state data includes: heart rate, blood pressure, body temperature, and respiratory rate; the movement data includes gait, posture, and exercise intensity; the location information data includes temperature, humidity, and altitude; and the bioelectrical signal sensing data includes skin conductance data and electromyography data.

[0030] Secondly, the present invention also provides an AI-based personalized pain management method, implemented using the AI-based personalized pain management system described above, comprising:

[0031] Real-time acquisition of multi-dimensional monitoring data, including user physiological status data, motion data, location information data, and bioelectrical signal sensing data;

[0032] Feature extraction is performed on physiological state data and bioelectric signal sensing data, and combined with basic user information, the user's pain type and pain level are identified.

[0033] Based on pain type, pain level, and multidimensional monitoring data, time series analysis is performed to predict the user's pain trend within a preset time period.

[0034] Output a pain management assessment report for the user based on pain type, pain level, and pain trend.

[0035] Thirdly, the present invention also provides an electronic device, including a processor and a memory, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, it implements the AI-based personalized pain management method described in the above technical solution.

[0036] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the AI-based personalized pain management method described in the above technical solution.

[0037] Compared with existing technologies, the AI-based personalized pain management system and method proposed in this invention have the following advantages:

[0038] 1. Real-time, multi-dimensional pain monitoring: Through a multi-dimensional data acquisition module, combined with physiological data, motion data, location information and bioelectric signal sensing data, the system can comprehensively monitor the user's pain status, overcoming the limitations of traditional methods;

[0039] 2. Personalized Pain Type Recognition: The pain type recognition module can accurately identify the type and level of pain by analyzing the user's physiological state and electrical signal data, thereby providing a customized pain management solution for the individual.

[0040] 3. Pain trend prediction: Through time series analysis, combined with LSTM network and dynamic time warping algorithm, it is possible to predict the pain trend of users in the future, so as to make prevention and intervention in advance, thereby helping users manage pain and avoid excessive pain attacks.

[0041] 4. Multidimensional assessment and feedback mechanism: The system generates pain management reports regularly through the feedback assessment module, based on the type, level and predictive trend of pain, providing professional analysis and suggestions.

[0042] In summary, the solution system of this invention integrates multi-dimensional data collection, personalized analysis, and feedback mechanisms, resulting in higher accuracy, personalization, and real-time performance in pain management. Compared to traditional technologies, it can provide more comprehensive, accurate, and personalized pain management solutions, significantly improving users' quality of life and health management efficiency. Attached Figure Description

[0043] Figure 1 is a schematic diagram of the structure of the AI-based personalized pain management system provided by the present invention;

[0044] Figure 2 is a schematic diagram of the AI-based personalized pain management method provided by the present invention;

[0045] Figure 3 is a schematic diagram of the electronic device structure provided by the present invention. Detailed Implementation

[0046] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0047] Please refer to Figure 1. This embodiment provides an AI-based personalized pain management system 100, including:

[0048] The multidimensional data acquisition module 101 is used to collect multidimensional monitoring data in real time, including user physiological state data, motion data, location information data, and bioelectric signal sensing data.

[0049] The pain type recognition module 102 is used to extract features from physiological state data and bioelectric signal sensing data, and identify the user's pain type and pain level by combining the user's basic information.

[0050] The pain trend prediction module 103 is used to perform time series analysis based on pain type, pain level and multidimensional monitoring data to predict the user's pain trend within a preset time period in the future.

[0051] The feedback assessment module 104 is used to output a pain management assessment report for the user based on the pain type, pain level, and pain trend.

[0052] The AI-based personalized pain management system provided in this embodiment comprehensively monitors a user's pain state through a multi-dimensional data acquisition module that combines physiological data, motion data, location information, and bioelectrical signal sensing data. By analyzing the user's physiological state and bioelectrical signal data, the system can accurately identify the type and level of pain. Through time series analysis, combined with LSTM networks and dynamic time warping algorithms, it can predict the user's pain trend over a future period, allowing for proactive prevention and intervention, thereby helping patients manage pain and avoid excessive pain attacks. The feedback evaluation module regularly generates pain management reports based on the type, level, and predicted trend of pain, providing professional analysis and suggestions to help users and medical personnel understand the effectiveness of pain management and adjust treatment plans. This system not only provides personalized pain monitoring and management solutions but also, through intelligent analysis and prediction, provides early warnings and optimizes treatment measures, significantly improving the effectiveness of pain management and the user's quality of life.

[0053] In a preferred embodiment, the multidimensional data acquisition module acquires the user's multidimensional status data through a wearable sensing device. Specifically, the physiological status data includes: heart rate, blood pressure, body temperature, and respiratory rate, acquired through a wristband-type biosensor; the motion data, including gait, posture, and exercise intensity, is acquired using a triaxial accelerometer; the location information data includes temperature, humidity, and altitude, which can be acquired through GPS data; the bioelectrical signal sensing data includes skin conductance data and electromyography (EMG) data, where skin conductance data reflects the body's autonomic nervous activity, especially changes related to physiological responses such as emotion and stress; and EMG data is used for motion monitoring, muscle fatigue assessment, etc.

[0054] In a preferred embodiment, the step of extracting features from physiological state data and bioelectrical signal sensing data, and identifying the user's pain type and pain level by combining them with basic user information, includes:

[0055] Extract SDNN and LF / HF power ratio from physiological state data;

[0056] Stress response values ​​were extracted from bioelectric signal sensing data.

[0057] The SDNN, LF / HF power ratio, and stress response values ​​are used as feature vectors and input into a trained KNN classifier to determine the pain type.

[0058] Based on the type of pain, a linear regression method is used to output a pain level score.

[0059] As a specific implementation, the standard deviation of the SDNN and the LF / HF power ratio are extracted to assess heart rate variability. A larger SDNN indicates a stronger regulatory capacity of the autonomic nervous system; a smaller SDNN indicates a greater cardiac stress response, which may be associated with adverse states such as pain or stress. Combined with skin conductance data (GSR), the main sites of stress are identified, and the slope of the rise, k = ΔGSR / Δt, is used as the stress response value to determine the degree of stress response. Typically, in a painful state, the slope of the rise in GSR increases, indicating a higher level of stress or pain.

[0060] As a specific implementation, K-Nearest Neighbors (KNN) classifies pain by calculating the distance between feature vectors. In this embodiment, the SDNN, LF / HF ratio, and stress response values ​​are combined into a single feature vector that describes the user's physiological state and bioelectrical signals. The KNN classifier is trained based on historical feature vectors and identifies the user's pain type (e.g., neuropathic / inflammatory pain) based on real-time input data. For example, pain with a GSR mutation rate > 0.8 Hz and an HRV low-frequency / high-frequency (LF / HF) ratio < 0.5 is classified as neuropathic pain.

[0061] In addition to KNN networks, in some embodiments, if there is enough user data, the DBSCAN scheme can also be used to cluster pain types in order to accurately locate and identify the user's pain type.

[0062] As a specific example, the main steps in establishing a linear regression model are as follows:

[0063] The first step is to collect training data: collect data with pain type labels and pain level scores, with pain type as the independent variable and pain level score as the dependent variable.

[0064] The second step is to input the pain type predicted by the KNN classifier, as well as other relevant features that may affect the pain level, into the model.

[0065] The third step is the training process: a linear regression model is trained using the collected data to predict pain levels. The quantitative numerical result used to predict pain is the pain level score.

[0066] In a preferred embodiment, the step of performing time series analysis based on pain type, pain level, and multidimensional monitoring data to predict the user's pain trend within a preset time period includes:

[0067] Multidimensional monitoring data is divided into time series data and static feature data;

[0068] Dynamic time warping is used to align time steps in time series data to construct a multidimensional feature vector.

[0069] Static feature data and multidimensional feature vectors are input into a prediction model based on an LSTM network for time series analysis to determine the pain level of users in each pain category within a preset future time period.

[0070] In a preferred embodiment, the structure of the prediction model based on the LSTM network is as follows:

[0071] An attention layer is added after the output of the base LSTM layer to dynamically assign weights to the output at each time step;

[0072] The static features and LSTM output are concatenated through a fully connected layer, and the concatenated data is mapped to the predicted pain score.

[0073] The model's loss function is the weighted mean squared error, and L2 regularization constraints are introduced.

[0074] Specifically, the static feature data includes user personal information (age, gender, BMI, medical history, etc.); the time series data includes signals such as heart rate and GSR that change over time.

[0075] Adding temporal attention to the LSTM network: Calculating the weights of key pain events at 6h / 12h / 24h, and the temporal attention weight vector α. t The calculation method for α is as follows: t =softmax(W q h t ·W k H T )

[0076] In the formula, W q W represents the query weight matrix, used to map the features of the current time step to the query space.k The key weight matrix represents the matrix used to map features from historical time steps to the key space; h t The pain-related features at time t are characterized; H contains all hidden states from the initial time to the current time, H = [h1, h2, ..., ht-1].

[0077] By simultaneously capturing immediate physiological responses and long-term behavioral patterns through an attention mechanism, the model can focus on key pain events at different time points, optimizing its ability to perceive both short-term and long-term events and calculating the weights of key pain events. This approach helps the model extract the most valuable information for prediction from global time-series data, capturing the combination of immediate physiological responses and long-term behavioral patterns, thereby improving prediction accuracy.

[0078] As a specific implementation, in this prediction network, we also introduce a residual connection method, introducing skip connections between stacked LSTM layers to alleviate gradient vanishing. This can be expressed as:

[0079] In the formula, This is used to directly transfer the basic features extracted by shallow networks (such as time-frequency features of physiological signals) to deep layers to prevent gradient vanishing; For skip connections, used to inherit the hidden state from the previous time step in this layer, capturing long-term patterns in pain evolution, x t These are the multimodal observations at the current moment. This represents the output of this LSTM layer.

[0080] By directly transferring features between shallow and deep networks, residual connections ensure that gradients can propagate effectively in multi-layer LSTM networks, thereby maintaining model stability during training and accelerating convergence.

[0081] In a preferred embodiment, the pain type identification module is deployed on a local server, enabling real-time processing of user data and rapid feedback of identification results. The pain trend prediction module is deployed on a cloud server, leveraging the powerful processing capabilities of cloud computing to perform more complex analyses and predictions. Furthermore, knowledge distillation is used to compress the cloud model into a Mobile-LSTM format, achieving model lightweighting. Additionally, the user terminal can enable a publish / subscribe mode based on the NATS messaging system: through the NATS publishing / subscribe mechanism, the user terminal can push pain data to the cloud for processing in real time, and the prediction module, acting as a subscriber, can receive and process the pushed data in real time. This mechanism ensures the speed and real-time nature of data flow, guaranteeing that the system can respond instantly to changes in the user's health status.

[0082] In a preferred embodiment, the output user's pain management assessment report includes:

[0083] The prediction results are dynamically displayed using heatmaps and dynamic trend charts;

[0084] If the level of a preset pain type exceeds the risk threshold within a future time period, the user's multidimensional monitoring data will be uploaded to the medical institution.

[0085] For example, in predicting pain in patients with chronic arthritis, the system uses historical data and current monitoring information to predict that the patient's joint pain intensity will continue to increase over the next 24 hours. A heatmap is used to display the patient's pain changes at different times of the day over the next few days. For instance, the heatmap might show that the patient's pain level is higher between 8 PM and 10 PM, with the color changing from yellow to red; simultaneously, a dynamic trend graph displays the curve of the patient's joint pain intensity changing over time. The prediction results are displayed to users via a mini-program.

[0086] If the prediction results indicate that the patient's pain intensity exceeds 8 points, accompanied by changes in physiological indicators (such as an increase in heart rate to 100 beats per minute), the system automatically uploads the patient's real-time monitoring data to the medical institution's server. After receiving the alert, the doctor may decide whether to adjust the treatment plan or proactively contact the user based on this information.

[0087] This invention also provides an AI-based personalized pain management method, implemented using any of the AI-based personalized pain management systems described above, comprising:

[0088] Step S101: Real-time acquisition of multi-dimensional monitoring data, including user physiological state data, motion data, location information data, and bioelectrical signal sensing data;

[0089] Step S102: Extract features from physiological state data and bioelectric signal sensing data, and identify the user's pain type and pain level by combining the user's basic information;

[0090] Step S103: Based on pain type, pain level and multidimensional monitoring data, perform time series analysis to predict the user's pain trend within a preset time period in the future;

[0091] Step S104: Output the user's pain management assessment report based on the pain type, pain level, and pain trend.

[0092] As shown in Figure 3, in addition to the aforementioned AI-based personalized pain management method, this invention also provides an electronic device 300, which can be a mobile terminal, desktop computer, laptop, handheld computer, server, or other computing device. The electronic device includes a processor 301, a memory 302, and a display 303.

[0093] In some embodiments, memory 302 may be an internal storage unit of a computer device, such as a hard disk or memory. In other embodiments, memory 302 may be an external storage device of a computer device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. Furthermore, memory 302 may include both internal and external storage units of the computer device. Memory 302 is used to store application software and various types of data installed on the computer device, such as program code for installing the computer device. Memory 302 may also be used to temporarily store data that has been output or will be output. In one embodiment, memory 302 stores an AI-based personalized pain management method program 304, which can be executed by processor 301 to implement an AI-based personalized pain management method according to various embodiments of the present invention.

[0094] In some embodiments, processor 301 may be a central processing unit (CPU), microprocessor, or other data processing chip, used to run program code stored in memory 302 or process data, such as executing an AI-based personalized pain management method program.

[0095] In some embodiments, display 303 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. Display 303 is used to display information on the computer device and to display a visual user interface. Components 301-303 of the computer device communicate with each other via a system bus.

[0096] This embodiment also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the AI-based personalized pain management method described in any of the above technical solutions.

[0097] The computer-readable storage medium and computing device provided in the above embodiments of the present invention can be implemented with reference to the content specifically described above for an AI-based personalized pain management method, and have similar beneficial effects to the AI-based personalized pain management method described above, which will not be repeated here.

[0098] This invention provides an AI-based personalized pain management system and method that can accurately identify different types and levels of pain, offering tailored pain management solutions for each user to maximize individualized needs. Based on time series analysis using LSTM networks and dynamic time warping algorithms, the system can not only identify the current pain status but also accurately predict pain trends over a future period, helping users take preventative and intervention measures before pain occurs, thus reducing suffering. Compared to traditional solutions, this invention provides a more comprehensive, accurate, and personalized pain management solution, significantly improving users' quality of life and health management efficiency.

[0099] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. An AI-based personalized pain management system, characterized by, include: The multidimensional data acquisition module is used to collect multidimensional monitoring data in real time, including user physiological state data, motion data, location information data, and bioelectrical signal sensing data. The pain type recognition module is used to extract features from physiological state data and bioelectric signal sensing data, and combine them with the user's basic information to identify the user's pain type and pain level. The pain trend prediction module is used to perform time series analysis based on pain type, pain level and multidimensional monitoring data to predict the user's pain trend within a preset time period in the future; The feedback assessment module is used to output a pain management assessment report for users based on pain type, pain level, and pain trend.

2. The AI-based personalized pain management system of claim 1, wherein, The process of extracting features from physiological state data and bioelectrical signal sensing data, and combining this with basic user information to identify the user's pain type and pain level, includes: Extract SDNN and LF / HF power ratio from physiological state data; Stress response values ​​were extracted from bioelectric signal sensing data. The SDNN, LF / HF power ratio, and stress response values ​​are used as feature vectors and input into a trained KNN classifier to determine the pain type. Based on the type of pain, a linear regression method is used to output a pain level score.

3. The AI-based personalized pain management system of claim 1, wherein, The method involves performing time series analysis based on pain type, pain level, and multidimensional monitoring data to predict the user's pain trend within a preset time period, including: Multidimensional monitoring data is divided into time series data and static feature data; Dynamic time warping is used to align time steps in time series data to construct a multidimensional feature vector. Static feature data and multidimensional feature vectors are input into a prediction model based on an LSTM network for time series analysis to determine the pain level of users in each pain category within a preset future time period.

4. The AI-based personalized pain management system of claim 3, wherein, The structure of the prediction model based on the LSTM network is as follows: An attention layer is added after the output of the base LSTM layer to dynamically assign weights to the output at each time step; The static features and LSTM output are concatenated through a fully connected layer, and the concatenated data is mapped to the predicted pain score. The model's loss function is the weighted mean squared error, and L2 regularization constraints are introduced.

5. The AI-based personalized pain management system of claim 1, wherein, The pain type identification module is deployed on a local server, and the pain trend prediction module is deployed on a cloud server. The model compression is achieved through knowledge distillation.

6. The AI-based personalized pain management system of claim 1, wherein, The output user's pain management assessment report includes: The prediction results are dynamically displayed using heatmaps and dynamic trend charts; If the level of a preset pain type exceeds the risk threshold within a future time period, the user's multidimensional monitoring data will be uploaded to the medical institution.

7. The AI-based personalized pain management system of any one of claims 1-6, wherein, Its features are, The physiological data includes: heart rate, blood pressure, body temperature, and respiratory rate; the exercise data includes gait, posture, and exercise intensity; the location information data includes temperature, humidity, and altitude; and the bioelectrical signal sensing data includes skin conductance data and electromyography data.

8. An AI-based personalized pain management method, characterized by, include: Real-time acquisition of multi-dimensional monitoring data, including user physiological status data, motion data, location information data, and bioelectrical signal sensing data; Feature extraction is performed on physiological state data and bioelectric signal sensing data, and combined with basic user information, the user's pain type and pain level are identified. Based on pain type, pain level, and multidimensional monitoring data, time series analysis is performed to predict the user's pain trend within a preset time period. Output a pain management assessment report for the user based on pain type, pain level, and pain trend.

9. An electronic device, comprising: It includes a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, it implements the AI-based personalized pain management method as described in claim 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the AI-based personalized pain management method as described in claim 8.

Images