Pain assessment and intervention system

Abstract

The invention discloses a pain assessment and intervention system, which belongs to the technical field of medical detection and comprises a user interaction module, a monitoring feedback module, a privacy security module, a cache management module, an assessment adjustment module, a comprehensive atlas module, a deep assessment module, an intervention suggestion module and a psychological support module. According to the method, more personalized pain assessment can be realized, the pain problem of a patient can be better managed, the workload of a doctor can be relieved, medical resources can be more reasonably distributed, the comprehensiveness and accuracy of pain assessment are improved, a more complete data basis is provided for comprehensive assessment, a deeper diagnosis basis is provided for the doctor, and the patient experience is improved. And the information related to the pain can be more comprehensively mined.

Description

Technical Field

[0001] This invention relates to the field of medical testing technology, and in particular to a pain assessment and intervention system. Background Technology

[0002] With societal development and advancements in medical technology, chronic diseases, especially knee osteoarthritis, have become a global focus of healthcare systems. Knee osteoarthritis is one of the most common chronic diseases, often accompanied by persistent and intractable pain, significantly impacting patients' lives. Pain is a primary driver of medical treatment and a major contributor to functional limitations and decreased quality of life. Furthermore, pain may be one of the most important variables in deciding whether to undergo surgery. However, current pain management models still face a number of challenges, including subjective assessment and difficulties in quantifying intervention effects.

[0003] Existing pain assessment and intervention systems cannot achieve more personalized pain assessment, and medical resources cannot be allocated rationally. In addition, the comprehensiveness and accuracy of pain assessment in existing pain assessment and intervention systems are reduced, and they cannot provide doctors with more in-depth diagnostic evidence. Therefore, we propose a pain assessment and intervention system. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a pain assessment and intervention system.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A pain assessment and intervention system includes a user interaction module, a monitoring and feedback module, a privacy and security module, a cache management module, an assessment and adjustment module, a comprehensive graph module, an in-depth assessment module, an intervention suggestion module, and a psychological support module;

[0007] The user interaction module is used to support interaction between patients and medical staff and to display visual information;

[0008] The monitoring feedback module is used to monitor and process patient data in real time;

[0009] The privacy and security module is used to encrypt and protect patient data during data collection and transmission.

[0010] The cache management module is used to collect and dynamically manage data cache information in the system.

[0011] The assessment and adjustment module is used to dynamically adjust the assessment strategy and select the optimal pain assessment model.

[0012] The integrated atlas module is used to construct a comprehensive medical atlas and provide an understanding of the patient's disease background.

[0013] The in-depth assessment module is used to analyze patient data through a pain assessment model and output an objective and comprehensive pain score.

[0014] The intervention suggestion module is used to generate personalized treatment suggestions based on pain assessment results;

[0015] The psychological support module is used to provide corresponding medical knowledge through a comprehensive medical atlas.

[0016] As a further aspect of the present invention, the specific steps for patient data processing in the monitoring feedback module are as follows:

[0017] Step 1: Detect and delete duplicate records, construct a filter window, select multiple adjacent data points within the filter window for each group of patient data, sort the selected data within the window, determine the value at the middle position, and then replace the current data point with the determined median value.

[0018] Step 2: Repeat the median selection and data replacement steps to iterate over the subsequent real-time patient data until each group of data is replaced by the sorted median within the window, and treat the incomplete window data as a cycle, filtering the beginning and end of the wave data;

[0019] Step 3: Calculate the mean and standard deviation of the collected patient data, and then obtain the threshold range of outliers based on the calculated mean and standard deviation to judge each group of data collected;

[0020] Step 4: If a patient's data exceeds the upper threshold or falls below the lower threshold, it is marked as an outlier. The marked outliers are then processed by either deleting or replacing them with the mean. After the outlier processing is complete, interpolation is used to fill in the missing values ​​in the patient data.

[0021] As a further aspect of the present invention, the specific calculation formula for the outlier threshold range in step three is as follows:

[0022]

[0023]

[0024] Z 上限 =μ+3×σ (3)

[0026] Z 下限 =μ-3×σ (3)

[0028] In the formula, μ represents the mean of the patient dataset; Xi σ represents each data point in the dataset; n represents the total number of data points; σ represents the standard deviation of the patient dataset.

[0029] As a further aspect of the present invention, the specific steps of the dynamic management of the cache management module are as follows:

[0030] Step 1: When the system starts, initialize a set of LRU linked lists and a hash table, and store patient data in the LRU linked lists according to the collection time from newest to oldest. When new data is accessed, check whether the data exists in the hash table.

[0031] Step 2: If it exists, it means that the data is already in the cache. Move the data node to the head of the LRU list to indicate that it has been accessed recently. If it does not exist, it means that the data is not in the cache.

[0032] Step 3: When data is detected not in the cache, remove the data node from the tail of the LRU list, delete the corresponding entry from the hash table, insert the newly accessed data node into the head of the LRU list, and insert a new entry into the hash table. After each data access, move the accessed data node to the head of the list.

[0033] As a further aspect of the present invention, the specific steps for dynamically adjusting the evaluation strategy of the evaluation adjustment module are as follows:

[0034] Step 1: The evaluation and adjustment module initializes a set of MCTS trees, sets the initial evaluation strategy and model set, and uses it as the root node of the MCTS tree. Starting from the root node of the tree, child nodes are selected according to the UCB1 strategy.

[0035] Step II: If the selected child node has been expanded, continue selecting until a child node that has not been fully expanded is selected and then stop. Use prior knowledge, experience or existing models to expand the selected node and generate possible next state.

[0036] Step 3: Based on real-time patient data, calculate the pain score of the node using a preset pain assessment model, send the simulated score back to the node in the MCTS tree, and update the cumulative score and access count of each group of nodes.

[0037] Step IV: Based on the backpropagation results, dynamically adjust the strategy for selecting child nodes. Based on the simulation and evaluation results, select the current optimal pain assessment model. Repeat the process of simulation, expansion, evaluation, backpropagation, strategy adjustment, and model selection until the predetermined number of iterations or time is reached, and then output the final pain assessment model. After that, the patient's pain level is assessed using this model.

[0038] As a further aspect of the present invention, the specific steps for constructing the comprehensive medical atlas using the comprehensive atlas module are as follows:

[0039] Step 1: Extract medical knowledge from medical literature, clinical practice and professional databases, then use named entity recognition algorithm to identify entities in the extracted text, identify medical entities, and link them to nodes in the knowledge graph;

[0040] Step 2: Using natural language processing technology, extract the relationships between medical entities, construct the edges in the knowledge graph, use a graph database to store the extracted medical knowledge in the form of a graph, periodically introduce new medical knowledge to expand and update the knowledge graph, and process the semantic relationships between entities in the knowledge graph.

[0041] As a further aspect of the present invention, the specific steps of patient data analysis in the in-depth assessment module are as follows:

[0042] Step 1: Collect historical patient data from the cache management module, preprocess the historical patient data, and then divide the preprocessed data into training set, validation set and test set;

[0043] Step 2: Construct a pain assessment model and input the training set into the pain assessment model. In each training iteration, input patient data into the pain assessment model, calculate the predicted value of the training set through forward pass, and calculate the loss through the cross-entropy loss function.

[0044] Step 3: Calculate the gradient of the loss with respect to the model parameters using the backpropagation algorithm, then clip the calculated gradient, and finally update the model's weights and biases using the SGD optimizer based on the calculated gradient.

[0045] Step 4: After each training round, validate the model using the validation set, adjust the network structure or hyperparameters, and repeat the training until the pre-set number of training rounds is reached or the training is stopped. Then, evaluate the model's performance using the test set and calculate the model's performance metrics by comparing the predicted pain level with the true labels.

[0046] Step 5: Use a weighted average method to integrate the output of the pain assessment model with the output of the assessment adjustment module to obtain a comprehensive pain score, and use the integrated comprehensive pain score as the objective assessment result.

[0047] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0048] 1. This invention initializes a set of MCTS trees and sets an initial assessment strategy and model set, which are then used as the root node of the MCTS trees. Starting from the root node, child nodes are selected according to the UCB1 strategy. If the selected child node has been expanded, the selection continues until a child node that has not been fully expanded is selected. Prior knowledge, experience, or existing models are used to expand the selected nodes, generating possible next-step states. Based on real-time patient data, the pain score of the node is calculated using a preset pain assessment model. The simulated score is fed back to the node of the MCTS tree, and the cumulative score and access count of each set of nodes are updated. Based on the backpropagation results, the strategy for selecting child nodes is dynamically adjusted. Based on the simulation and assessment results, the current optimal pain assessment model is selected. The process of simulation, expansion, assessment, backpropagation, strategy adjustment, and model selection is repeated until a predetermined number of iterations or time is reached, at which point the final pain assessment model is output. This model is then used to assess the patient's pain level, which helps to achieve more personalized pain assessment, better manage patients' pain problems, reduce the workload of doctors, and make medical resources more rationally allocated.

[0049] 2. This invention collects and preprocesses historical patient data, then divides the preprocessed data into training, validation, and test sets to construct a pain assessment model. The training set is input into the pain assessment model. In each training iteration, patient data is input into the pain assessment model. The predicted values ​​of the training set are calculated through forward propagation, the loss is calculated using the cross-entropy loss function, and the gradient of the loss with respect to the model parameters is calculated using the backpropagation algorithm. The calculated gradient is then clipped. Based on the calculated gradient, the SGD optimizer is used to update the model's weights and biases. After each round of training, the model is validated using the validation set. The entire network structure or hyperparameters are repeatedly trained until a pre-set number of training rounds is reached or training is stopped. The performance of the model is then evaluated using a test set. The model's performance index is calculated by comparing the predicted pain level with the true labels. A weighted average is used to integrate the output of the pain assessment model with the output of the assessment adjustment module to obtain a comprehensive pain score. The integrated comprehensive pain score is then used as an objective evaluation result, improving the comprehensiveness and accuracy of pain assessment. This provides a more complete data foundation for comprehensive assessment, offers doctors a deeper diagnostic basis, and enables a more comprehensive mining of pain-related information. Attached Figure Description

[0050] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0051] Figure 1This is a system block diagram of a pain assessment and intervention system proposed in this invention. Detailed Implementation

[0052] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0053] Example 1

[0054] Reference Figure 1 A pain assessment and intervention system includes a user interaction module, a monitoring and feedback module, a privacy and security module, a cache management module, an assessment and adjustment module, a comprehensive graph module, an in-depth assessment module, an intervention suggestion module, and a psychological support module.

[0055] The user interaction module supports interaction between patients and medical staff and displays visual information; the monitoring and feedback module is used to monitor and process patient data in real time.

[0056] Specifically, duplicate records are detected and deleted. A filtering window is constructed. For each group of patient data, multiple adjacent data points are selected within the filtering window. The selected data within the window are then sorted, and the median value is determined. The determined median value replaces the current data point. This median selection and data replacement process is repeated to iterate over subsequent real-time patient data until each group of data is replaced by the sorted median value within the window. Incomplete window data is considered as a cycle, and filtering is performed at the beginning and end of the data. The mean and standard deviation of the collected patient data are calculated. Then, based on the calculated mean and standard deviation, a threshold range for outliers is obtained. Each group of collected data is judged. If a patient data point exceeds the upper limit of the threshold or is less than the lower limit, it is marked as an outlier. The marked outliers are then processed by deleting or replacing them with the mean. After outlier processing, interpolation is used to fill in missing values ​​in the patient data.

[0057] In this embodiment, the specific formula for calculating the outlier threshold range is as follows:

[0058]

[0059]

[0060] Z 上限 =μ+3×σ (3)

[0062] Z 下限 =μ-3×σ (3)

[0064] In the formula, μ represents the mean of the patient dataset; Xi σ represents each data point in the dataset; n represents the total number of data points; σ represents the standard deviation of the patient dataset.

[0065] The privacy and security module is used to encrypt and protect patient data during data collection and transmission; the cache management module is used to collect and dynamically manage data cache information in the system.

[0066] Specifically, when the system starts, an LRU linked list and a hash table are initialized. Patient data is stored in the LRU linked list from newest to oldest according to the collection time. When new data is accessed, it is checked whether the data exists in the hash table. If it exists, it means that the data is already in the cache, and the data node is moved to the head of the LRU linked list to indicate that it has been accessed recently. If it does not exist, it means that the data is not in the cache. When it is detected that the data is not in the cache, the data node is taken from the tail of the LRU linked list, the corresponding entry is deleted from the hash table, the newly accessed data node is inserted into the head of the LRU linked list, and a new entry is inserted into the hash table. After each data access, the accessed data node is moved to the head of the linked list.

[0067] The assessment and adjustment module is used to dynamically adjust the assessment strategy and select the optimal pain assessment model.

[0068] Specifically, the assessment and adjustment module initializes a set of MCTS trees and sets an initial assessment strategy and model set, which is then used as the root node of the MCTS trees. Starting from the root node, child nodes are selected according to the UCB1 strategy. If the selected child node has been expanded, the selection continues until a child node that has not been fully expanded is selected. Prior knowledge, experience, or existing models are used to expand the selected nodes, generating possible next states. Based on real-time patient data, the pain score of the node is calculated using a preset pain assessment model. The simulated score is fed back to the node of the MCTS tree, and the cumulative score and access count of each set of nodes are updated. Based on the backpropagation results, the strategy for selecting child nodes is dynamically adjusted. Based on the simulation and assessment results, the current optimal pain assessment model is selected. The process of simulation, expansion, assessment, backpropagation, strategy adjustment, and model selection is repeated until the predetermined number of iterations or time is reached, at which point the final pain assessment model is output. The patient's pain level is then assessed using this model.

[0069] Example 2

[0070] Reference Figure 1 A pain assessment and intervention system includes a user interaction module, a monitoring and feedback module, a privacy and security module, a cache management module, an assessment and adjustment module, a comprehensive graph module, an in-depth assessment module, an intervention suggestion module, and a psychological support module.

[0071] The comprehensive atlas module is used to construct a comprehensive medical atlas and provide an understanding of the patient's disease background.

[0072] Specifically, medical knowledge is extracted from medical literature, clinical practice, and professional databases. Named entity recognition algorithms are then used to identify entities in the extracted text, and medical entities are identified and linked to nodes in the knowledge graph. Natural language processing techniques are used to extract the relationships between medical entities, construct edges in the knowledge graph, and store the extracted medical knowledge in the form of a graph database. New medical knowledge is periodically introduced to expand and update the knowledge graph, and the semantic relationships between entities in the knowledge graph are processed.

[0073] The in-depth assessment module is used to analyze patient data through a pain assessment model and output an objective and comprehensive pain score.

[0074] Specifically, historical patient data is collected from the cache management module, preprocessed, and then divided into training, validation, and test sets to construct a pain assessment model. The training set is input into the pain assessment model. In each training iteration, patient data is input into the pain assessment model, and the predicted values ​​of the training set are calculated through forward propagation. The loss is calculated using the cross-entropy loss function, and the gradient of the loss with respect to the model parameters is calculated using the backpropagation algorithm. The calculated gradient is then clipped. Based on the calculated gradient, the SGD optimizer is used to update the model's weights and biases. After each training round, the model is validated using the validation set, and the network structure or hyperparameters are adjusted. Training is repeated until a pre-set number of training rounds is reached or the training stops. The model's performance is then evaluated using the test set. The model's performance index is calculated by comparing the predicted pain level with the true label. A weighted average is used to integrate the output of the pain assessment model with the output of the evaluation adjustment module to obtain a comprehensive pain score. The integrated comprehensive pain score is then used as the objective evaluation result.

[0075] The intervention suggestion module generates personalized treatment recommendations based on pain assessment results; the psychological support module provides corresponding medical knowledge through a comprehensive medical atlas.

Claims

1. A pain assessment and intervention system, characterized by, It includes a user interaction module, a monitoring and feedback module, a privacy and security module, a cache management module, an assessment and adjustment module, a comprehensive graph module, an in-depth assessment module, an intervention suggestion module, and a psychological support module; The user interaction module is used to support interaction between patients and medical staff and to display visual information; The monitoring feedback module is used to monitor and process patient data in real time; The privacy and security module is used to encrypt and protect patient data during data collection and transmission. The cache management module is used to collect and dynamically manage data cache information in the system. The assessment and adjustment module is used to dynamically adjust the assessment strategy and select the optimal pain assessment model. The integrated atlas module is used to construct a comprehensive medical atlas and provide an understanding of the patient's disease background. The in-depth assessment module is used to analyze patient data through a pain assessment model and output an objective and comprehensive pain score. The intervention suggestion module is used to generate personalized treatment suggestions based on pain assessment results; The psychological support module is used to provide corresponding medical knowledge through a comprehensive medical atlas.

2. A pain assessment and intervention system according to claim 1, wherein, The specific steps for patient data processing in the monitoring and feedback module are as follows: Step 1: Detect and delete duplicate records, construct a filter window, select multiple adjacent data points within the filter window for each group of patient data, sort the selected data within the window, determine the value at the middle position, and then replace the current data point with the determined median value. Step 2: Repeat the median selection and data replacement steps to iterate over the subsequent real-time patient data until each group of data is replaced by the sorted median within the window, and treat the incomplete window data as a cycle, filtering the beginning and end of the wave data; Step 3: Calculate the mean and standard deviation of the collected patient data, and then obtain the threshold range of outliers based on the calculated mean and standard deviation to judge each group of data collected; Step 4: If a patient's data exceeds the upper threshold or falls below the lower threshold, it is marked as an outlier. The marked outliers are then processed by either deleting or replacing them with the mean. After the outlier processing is complete, interpolation is used to fill in the missing values ​​in the patient data.

3. A pain assessment and intervention system according to claim 2, wherein, The specific calculation formula for the outlier threshold range mentioned in step three is as follows: Z 上限 = μ + 3 × σ (3) Z 下限 = μ - 3 × σ (3) where μ represents the mean of the patient data set; X i represents each set of data points in the data set; n represents the total number of data points; and σ represents the standard deviation of the patient data set.

4. The pain assessment and intervention system of claim 2, wherein, The specific steps for dynamic management by the cache management module are as follows: Step 1: When the system starts, initialize a set of LRU linked lists and a hash table, and store patient data in the LRU linked lists according to the collection time from newest to oldest. When new data is accessed, check whether the data exists in the hash table. Step 2: If it exists, it means that the data is already in the cache. Move the data node to the head of the LRU list to indicate that it has been accessed recently. If it does not exist, it means that the data is not in the cache. Step 3: When data is detected not in the cache, remove the data node from the tail of the LRU list, delete the corresponding entry from the hash table, insert the newly accessed data node into the head of the LRU list, and insert a new entry into the hash table. After each data access, move the accessed data node to the head of the list.

5. A pain assessment and intervention system according to claim 4, wherein, The specific steps for dynamically adjusting the evaluation strategy in the evaluation adjustment module are as follows: Step 1: The evaluation and adjustment module initializes a set of MCTS trees, sets the initial evaluation strategy and model set, and uses it as the root node of the MCTS tree. Starting from the root node of the tree, child nodes are selected according to the UCB1 strategy. Step II: If the selected child node has been expanded, continue selecting until a child node that has not been fully expanded is selected and then stop. Use prior knowledge, experience or existing models to expand the selected node and generate possible next state. Step 3: Based on real-time patient data, calculate the pain score of the node using a preset pain assessment model, send the simulated score back to the node in the MCTS tree, and update the cumulative score and access count of each group of nodes. Step IV: Based on the backpropagation results, dynamically adjust the strategy for selecting child nodes. Based on the simulation and evaluation results, select the current optimal pain assessment model. Repeat the process of simulation, expansion, evaluation, backpropagation, strategy adjustment, and model selection until the predetermined number of iterations or time is reached, and then output the final pain assessment model. After that, the patient's pain level is assessed using this model.

6. The pain assessment and intervention system of claim 1, wherein, The specific steps for constructing the comprehensive medical atlas module are as follows: Step 1: Extract medical knowledge from medical literature, clinical practice and professional databases, then use named entity recognition algorithm to identify entities in the extracted text, identify medical entities, and link them to nodes in the knowledge graph; Step 2: Using natural language processing technology, extract the relationships between medical entities, construct the edges in the knowledge graph, use a graph database to store the extracted medical knowledge in the form of a graph, periodically introduce new medical knowledge to expand and update the knowledge graph, and process the semantic relationships between entities in the knowledge graph.

7. A pain assessment and intervention system according to claim 5, wherein, The specific steps for patient data analysis in the in-depth assessment module are as follows: Step 1: Collect historical patient data from the cache management module, preprocess the historical patient data, and then divide the preprocessed data into training set, validation set and test set; Step 2: Construct a pain assessment model and input the training set into the pain assessment model. In each training iteration, input patient data into the pain assessment model, calculate the predicted value of the training set through forward pass, and calculate the loss through the cross-entropy loss function. Step 3: Calculate the gradient of the loss with respect to the model parameters using the backpropagation algorithm, then clip the calculated gradient, and finally update the model's weights and biases using the SGD optimizer based on the calculated gradient. Step 4: After each training round, validate the model using the validation set, adjust the network structure or hyperparameters, and repeat the training until the pre-set number of training rounds is reached or the training is stopped. Then, evaluate the model's performance using the test set and calculate the model's performance metrics by comparing the predicted pain level with the true labels. Step 5: Use a weighted average method to integrate the output of the pain assessment model with the output of the assessment adjustment module to obtain a comprehensive pain score, and use the integrated comprehensive pain score as the objective assessment result.

Images