Personalized postoperative pain management method and system for children

Abstract

The present invention discloses a personalized postoperative pain management method and system for children, and relates to the technical field of pain management. A personalized postoperative pain management system for children includes: a data acquisition module, a developmental assessment module, a sample management module, a pain assessment module, a scheme matching module, and a feedback adjustment module. The present invention combines objective and subjective assessments to achieve comprehensive collection and integration of multi-dimensional data, greatly improving the accuracy of pain assessment; through psychological development and physiological development stage assessments, weighted analysis of patients is performed to ensure that pain assessment results and management plans fully consider the individual characteristics of children, significantly improving the pertinence and scientificity of pain management.

Description

Technical Field

[0001] The present invention relates to the technical field of pain management, and in particular to a personalized method and system for postoperative pain management in children. Background Art

[0002] In the management of postoperative pain in children, traditional methods mostly rely on the subjective judgment of doctors and a single treatment method, which is difficult to meet the personalized needs of children's pain management. Children show different pain perception and reaction due to individual differences in their physiological and psychological development stages. It is difficult for existing technologies to fully integrate these characteristics. The main problems are as follows: First, pain assessment mostly relies on patients' subjective descriptions or single monitoring indicators (such as heart rate, blood pressure, etc.). In particular, children have low assessment accuracy due to language expression and cognitive limitations. At the same time, the impact of parents' anxiety levels and care behaviors on children's pain management has not been fully considered. Secondly, children's pain perception is significantly affected by their physiological and psychological development stages. Children of the same age show different pain tolerance and coping patterns due to developmental differences. Existing methods fail to effectively evaluate and use developmental stages as a basis for management. In addition, existing management plans lack personalized adjustments, mostly adopt a unified treatment model, ignore individual differences in children, and non-drug treatments (such as psychological intervention and dietary adjustments) are also not targeted. Finally, pain management is usually based on a single assessment result, lacks real-time monitoring and dynamic adjustment, and cannot continuously optimize treatment plans to meet the personalized needs of patients.

[0003] To address these issues, the present invention provides personalized and precise solutions for children's postoperative pain management by integrating objective and subjective assessment data, introducing developmental stage analysis, a pain assessment model based on deep learning, and a real-time feedback mechanism, thereby significantly improving recovery effects and quality of life. Summary of the invention

[0004] In order to solve the technical problems raised in the above background technology, the present invention provides a personalized method and system for postoperative pain management in children.

[0005] A personalized approach to postoperative pain management in children, including the following steps:

[0006] Step S1: Data collection, collecting pain assessment data D, the pain assessment data includes an objective assessment part and a subjective assessment part, wherein:

[0007] The objective assessment part includes patient identity information, surgical information, diet and daily routine information, and medical imaging information;

[0008] The subjective assessment part includes the patient part and the parent part. The patient part includes the patient's preoperative anxiety information, the patient's current pain assessment information and the patient's pain feedback imaging information. The parent part includes the parent's preoperative anxiety information and parent care information.

[0009] Step S2: Developmental stage assessment, conducting a developmental stage assessment on the patient, the developmental stage includes the psychological developmental stage PS and the physiological developmental stage PH, the developmental stage is determined based on age through a professional assessment tool, and the developmental stage assessment result is used to weight the patient portion of the pain assessment data and match the personalized pain management plan;

[0010] Step S3: sample construction, preprocessing the pain assessment data, and constructing a personalized patient sample based on the preprocessed pain assessment data and developmental stage;

[0011] Step S4: pain assessment, sending the personalized patient sample into the pain assessment model based on the hybrid deep neural network for calculation to obtain the pain assessment result, which includes primary parameters and secondary parameters, wherein the primary parameters include pain location, pain state, and pain degree, and the secondary parameters include pain urgency, pain level, and pain mechanism;

[0012] Step S5: matching a personalized pain management plan, matching a personalized pain management plan according to the pain assessment results and the developmental stage, wherein the personalized pain management plan includes drug therapy and non-drug therapy, and the non-drug therapy includes physical therapy, psychological therapy, diet therapy, and work and rest therapy;

[0013] Step S6: Real-time feedback and adjustment of pain management plan. According to the feedback data after the patient's pain management, the personalized patient sample is updated and the personalized pain management plan is adjusted in real time.

[0014] As a preferred technical solution of the present invention, the evaluation of the developmental stage includes the following two parts:

[0015] Psychological development stage assessment: The patient's psychological development stage is assessed through standardized psychological assessment tools to obtain the psychological development age. The psychological assessment tools include but are not limited to the Children's Anxiety Scale, Children's Pain Coping Style Scale and Behavior Observation Scale; the results of the psychological development stage assessment include the patient's anxiety level, emotional response and cognitive pattern of pain;

[0016] Physiological development stage assessment: The physiological development stage is assessed through the patient's physiological characteristics, which include age, height, weight and physiological indicators, combined with physiological development standards to determine the patient's physiological maturity; physiological assessment tools include but are not limited to age-weight curves and physiological growth standard curves.

[0017] As a preferred technical solution of the present invention, preprocessing the pain assessment data includes:

[0018] Data cleaning: remove invalid data, outliers and noise to ensure data accuracy;

[0019] Data standardization: converting data from different sources into standardized feature vectors so that the data can be used as input for the pain assessment model;

[0020] Weighted adjustment: The patient portion of the pain assessment data is weighted according to the child's developmental stage to ensure that the assessment results truly reflect the patient's overall condition;

[0021] The weighted adjustment is calculated as: , where is the weighted and adjusted pain assessment data for item x, is the pain assessment data for item x, and They are the values ​​for the psychological development stage and the physiological development stage, respectively.

[0022] As a preferred technical solution of the present invention, the pain assessment model includes:

[0023] The pain assessment model is based on a hybrid deep neural network, which is constructed by combining a convolutional neural network and a long short-term memory network. The convolutional neural network is used to extract features from the patient's medical imaging information and pain feedback imaging information. The long short-term memory network is used to process time series data and analyze the changing trends and feedback information of the patient's postoperative pain to capture the pain evolution pattern over a long time span.

[0024] As a preferred technical solution of the present invention, the structure of the pain assessment model includes:

[0025] An input layer, for receiving personalized patient samples, comprising two independent input gates, for inputting data selection in the personalized patient samples into the convolution hidden layer and the memory hidden layer respectively;

[0026] Hidden layer, used to extract deep features in the data, including convolution hidden layer, memory hidden layer and evaluation hidden layer. The convolution hidden layer and memory hidden layer correspond to the fully connected layer, convolution layer and pooling layer of the pre-trained convolutional neural network and long short-term memory network respectively; the evaluation hidden layer is used to comprehensively evaluate the outputs of the convolution hidden layer and the memory hidden layer;

[0027] The output layer generates pain assessment parameters based on the comprehensive assessment results;

[0028] The output of the pain assessment model is a multi-label classification, which is used to predict multiple pain assessment parameters, including pain location P l , pain status P s , pain level P d and pain relief u , pain level S a , pain mechanism m , among which the pain level The calculation formula is: , where and are the feature sets of the input convolutional neural network and the long short-term neural network, respectively. and are the training weights corresponding to the two models respectively. is the bias term for calculation;

[0029] The loss function of the pain assessment model uses a weighted cross entropy loss function to handle the problem of multi-class imbalance. The calculation formula is: , where represents the loss function value, represents the number of categories, Represents the category index, is the weight coefficient of the i-th category, is the true label of the i-th category, is the predicted probability of the i-th category;

[0030] The pain assessment model was optimized through transfer learning technology. Feature extraction was performed by pre-training convolutional neural networks and long short-term memory networks on the same sample set, and fine-tuning was performed based on the patient's personalized data to obtain a hybrid deep neural network for constructing the pain assessment model.

[0031] As a preferred technical solution of the present invention, matching a personalized pain management solution includes:

[0032] Combine the pain assessment results, the patient's developmental stage assessment results, and the patient's personal preferences and medical needs to match a preliminary personalized pain management plan. The plan design includes drug therapy and non-drug therapy. According to the complexity of pain and individual differences of patients, the preliminary plan is adjusted through multi-level optimization rules, including pain priority, patient development value and risk control value.

[0033] The personalized management plan is matched through the following formula: , where Indicates the final matching solution. , , are the adjustment coefficients for pain priority, patient development value, and risk control value, respectively. and Respectively represent the weighted sum of the main parameters and the weighted sum of the secondary parameters, Represents the patient's developmental value, which is used to adjust the personalized weight of the scheme matching, through the psychological development stage value and physiological development stage values The weighted sum of Represents the risk control value, which is used to constrain the management plan to ensure its safety, feasibility and patient acceptance. The calculation formula is as follows: , where is the number of constraints, represents the i-th constraint, Represents the weight corresponding to the i-th constraint.

[0034] As a preferred technical solution of the present invention, real-time feedback and pain management solution adjustment include:

[0035] As the treatment progresses, the real-time monitoring system continuously collects feedback data from patients. The feedback data includes pain intensity and emotional changes. Based on the feedback data, the personalized patient sample is updated in real time, and the pain assessment results are recalculated through the pain assessment model to adjust the personalized pain management plan.

[0036] A personalized postoperative pain management system for children, including the following modules:

[0037] A data collection module is used to collect pain assessment data, wherein the pain assessment data includes an objective assessment part and a subjective assessment part, wherein: the objective assessment part includes patient identity information, surgery information, diet and work and rest information, and medical imaging information; the subjective assessment part includes a patient part and a parent part, wherein the patient part includes the patient's preoperative anxiety information, the patient's current pain assessment information, and the patient's pain feedback imaging information, and the parent part includes the parent's preoperative anxiety information and the parent's care information; the data collection module includes but is not limited to a smart bracelet, a sensor, a facial expression recognition device, and a voice recognition device;

[0038] A developmental stage assessment module is used to conduct developmental stage assessment on patients, including a psychological assessment unit and a physiological assessment unit, which are respectively used to assess the psychological developmental stage PS and the physiological developmental stage PH, wherein the developmental stage is determined based on age through professional assessment tools, and the developmental stage assessment results are used to weight the patient portion of the pain assessment data and match the personalized pain management plan; a sample management module includes a data preprocessing unit, a sample updating unit and a sample management unit; the data preprocessing unit is used to preprocess the pain assessment data; the sample updating unit is used to construct a personalized patient sample based on the preprocessed pain assessment data and the developmental stage, and to update the personalized patient sample based on the feedback data; the sample management unit is used to actively adjust the personalized patient sample;

[0039] The pain assessment module is composed of a pain assessment model, which is used to send personalized patient samples into the pain assessment model based on a hybrid deep neural network for calculation to obtain pain assessment results, which include primary parameters and secondary parameters, wherein the primary parameters include pain location, pain state, and pain degree, and the secondary parameters include pain urgency, pain level, and pain mechanism;

[0040] A program matching module, used for matching a personalized pain management program according to the pain assessment results and the developmental stage, wherein the personalized pain management program includes drug therapy and non-drug therapy, and the non-drug therapy includes physical therapy, psychological therapy, diet therapy and work and rest therapy;

[0041] The feedback adjustment module is used to update personalized patient samples and adjust personalized pain management plans in real time based on the feedback data after patient pain management.

[0042] The present invention has the following advantages:

[0043] The present invention combines objective and subjective evaluations to achieve comprehensive collection and integration of multi-dimensional data, thereby greatly improving the accuracy of pain assessment. Through psychological development and physiological development stage assessments, weighted analysis of patients is performed to ensure that pain assessment results and management plans fully consider the individual characteristics of children, significantly improving the pertinence and scientific nature of pain management.

[0044] Through the hybrid deep neural network (combining convolutional neural network and long short-term memory network), the present invention can extract deep features from medical images and time series data, accurately predict children's pain status, degree and mechanism, and provide strong technical support for personalized management.

[0045] The present invention uses pain assessment results, combines drug treatment with non-drug treatment, achieves personalized matching of plans, provides diversified management methods, and optimizes the quality of postoperative recovery; collects patient feedback through a real-time monitoring system, and dynamically adjusts the personalized pain management plan in combination with the evaluation model to ensure that the treatment plan can respond to changes in patient pain in a timely manner and improve the treatment effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a schematic diagram of the structure of a personalized children's postoperative pain management system used in an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to enable persons skilled in the art to better understand the technical solutions in 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 in the embodiments of the present invention.

[0048] Embodiment 1, a personalized method for postoperative pain management in children, comprising the following steps:

[0049] Step S1: Data collection, collecting the patient's pain assessment data D, the pain assessment data includes an objective assessment part and a subjective assessment part, wherein:

[0050] The objective assessment part includes patient identity information, surgical information, diet and daily routine information, and medical imaging information;

[0051] The subjective assessment part includes the patient part and the parent part. The patient part includes the patient's preoperative anxiety information, the patient's current pain assessment information and the patient's pain feedback imaging information. The parent part includes the parent's preoperative anxiety information and parent care information.

[0052] Patient identity information, including the patient's name, gender, age, hospitalization number, surgical department, hospitalization time, and medical record number, is used to uniquely identify the patient and ensure the accuracy and traceability of data collection and management. For example, a 5-year-old boy A, hospitalization number 123456, underwent appendectomy.

[0053] Surgical information, including the name of the surgery (e.g., appendectomy), type (e.g., minimally invasive surgery), anesthesia method (e.g., general anesthesia), duration of surgery (e.g., 90 minutes), and postoperative analgesia measures (e.g., epidural analgesia pump), is used to analyze the possible sources and severity of postoperative pain. For example, patient A underwent minimally invasive appendectomy, with general anesthesia during the surgery and an epidural analgesia pump for continuous infusion of analgesics after the surgery.

[0054] Diet and daily routine information, including the time of the patient's three meals a day, the types of food consumed and the nutritional content, as well as the duration of sleep at night, sleep quality (such as the proportion of deep sleep), and daytime activity level. For example: On the third day after surgery, patient A mainly ate liquid food (rice porridge, juice), consumed a total of 1200 kcal per day, slept for about 8 hours, and deep sleep accounted for 50%.

[0055] Medical imaging information, including postoperative imaging examinations (such as postoperative abdominal CT showing the scope of inflammation) and imaging records of wound healing (taking wound photos through a camera). For example: Patient A’s CT scan on the second day after surgery showed that the appendix inflammation was basically under control, the wound healed well, and there was no obvious sign of infection.

[0056] Subjective data, patient part: preoperative anxiety was assessed by the Children's Anxiety Scale (such as A score of 5, indicating moderate anxiety); current pain was assessed by facial expression score (such as VAS score of 7), VAS is the visual analog score of pain; pain feedback imaging was recorded by real-time facial video of the patient, and pain-related behaviors (such as frowning, body twisting) were recorded. Parent part: preoperative anxiety of parents was scored as 6 (indicating high anxiety) by the anxiety assessment scale, and parent care behavior was recorded by interview, with an average daily nursing time of 6 hours.

[0057] Step S2: Developmental stage assessment, conducting a developmental stage assessment on the patient, the developmental stage includes the psychological developmental stage PS and the physiological developmental stage PH, the developmental stage is determined based on age through a professional assessment tool, and the developmental stage assessment result is used to weight the patient portion of the pain assessment data and match the personalized pain management plan;

[0058] The developmental assessment consists of two parts:

[0059] Psychological development stage assessment: The patient's psychological development stage is assessed through standardized psychological assessment tools to obtain the psychological development age. The psychological assessment tools include but are not limited to the Children's Anxiety Scale, Children's Pain Coping Style Scale and Behavior Observation Scale; the results of the psychological development stage assessment include the patient's anxiety level, emotional response and cognitive pattern of pain;

[0060] Physiological development stage assessment: The physiological development stage is assessed through the patient's physiological characteristics, which include age, height, weight and physiological indicators, combined with physiological development standards to determine the patient's physiological maturity; physiological assessment tools include but are not limited to age-weight curves and physiological growth standard curves.

[0061] For example: Assessment of psychological development stage, using the Child Anxiety Scale and Behavior Observation Scale to assess the psychological development stage. Patient A has a psychological age of 4 years old, a moderate level of anxiety, and a weak ability to cope with pain. He may need a pain management plan that focuses more on psychological support.

[0062] Physiological development stage assessment, combined with the patient's age (5 years old), height (110cm), weight (18kg), the physiological development stage was assessed through a standardized curve. Patient A's physiological development was consistent with the level of a normal 5-year-old child, and there were no abnormalities in physiological indicators (heart rate, blood pressure, etc.).

[0063] Step S3: sample construction, preprocessing the pain assessment data, and constructing a personalized patient sample based on the preprocessed pain assessment data and developmental stage;

[0064] Preprocessing of pain assessment data includes:

[0065] Data cleaning: Remove invalid data, outliers and noise to ensure data accuracy; for example, patient A’s activity data on the first day was lost due to equipment failure and was automatically removed by the system.

[0066] Data standardization converts data from various sources into standardized feature vectors so that the data can be input into the pain assessment model; the dietary intake of patient A (such as 1200 kcal per day) is standardized into a feature vector to ensure that the data is uniformly input into the model.

[0067] Weighted adjustment: The patient portion of the pain assessment data is weighted according to the child's developmental stage to ensure that the assessment results truly reflect the patient's overall condition;

[0068] The weighted adjustment is calculated as: , where is the weighted and adjusted pain assessment data for item x, is the pain assessment data for item x, and They are the values ​​for the psychological development stage and the physiological development stage, respectively.

[0069] Step S4: pain assessment, sending the personalized patient sample into the pain assessment model based on the hybrid deep neural network for calculation to obtain the pain assessment result, which includes primary parameters and secondary parameters, wherein the primary parameters include pain location, pain state, and pain degree, and the secondary parameters include pain urgency, pain level, and pain mechanism;

[0070] The pain assessment model includes:

[0071] The pain assessment model is based on a hybrid deep neural network, which is constructed by combining a convolutional neural network and a long short-term memory network. The convolutional neural network is used to extract features from the patient's medical imaging information and pain feedback imaging information. The long short-term memory network is used to process time series data and analyze the changing trends and feedback information of the patient's postoperative pain to capture the pain evolution pattern over a long time span.

[0072] The structure of the pain assessment model includes:

[0073] An input layer, for receiving personalized patient samples, comprising two independent input gates, for inputting data selection in the personalized patient samples into the convolution hidden layer and the memory hidden layer respectively;

[0074] Hidden layer, used to extract deep features in the data, including convolution hidden layer, memory hidden layer and evaluation hidden layer. The convolution hidden layer and memory hidden layer correspond to the fully connected layer, convolution layer and pooling layer of the pre-trained convolutional neural network and long short-term memory network respectively; the evaluation hidden layer is used to comprehensively evaluate the outputs of the convolution hidden layer and the memory hidden layer;

[0075] The output layer generates pain assessment parameters based on the comprehensive assessment results;

[0076] The output of the pain assessment model is a multi-label classification, which is used to predict multiple pain assessment parameters, including pain location P l , pain status P s , pain level P d and pain relief u , pain level Sa , pain mechanism m , among which the pain level The calculation formula is: , where and are the feature sets of the input convolutional neural network and the long short-term neural network, respectively. and are the training weights corresponding to the two models respectively. is the bias term for calculation;

[0077] The loss function of the pain assessment model uses a weighted cross entropy loss function to handle the problem of multi-class imbalance. The calculation formula is: , where represents the loss function value, represents the number of categories, Represents the category index, is the weight coefficient of the i-th category, is the true label of the i-th category, is the predicted probability of the i-th category;

[0078] The pain assessment model was optimized through transfer learning technology. Feature extraction was performed by pre-training convolutional neural networks and long short-term memory networks on the same sample set, and fine-tuning was performed based on the patient's personalized data to obtain a hybrid deep neural network for constructing the pain assessment model.

[0079] Step S5: matching a personalized pain management plan, matching a personalized pain management plan according to the pain assessment results and the developmental stage, wherein the personalized pain management plan includes drug therapy and non-drug therapy, and the non-drug therapy includes physical therapy, psychological therapy, diet therapy, and work and rest therapy;

[0080] Personalized pain management solutions include:

[0081] Combine the pain assessment results, the patient's developmental stage assessment results, and the patient's personal preferences and medical needs to match a preliminary personalized pain management plan. The plan design includes drug therapy and non-drug therapy. According to the complexity of pain and individual differences of patients, the preliminary plan is adjusted through multi-level optimization rules, including pain priority, patient development value and risk control value.

[0082] The personalized management plan is matched through the following formula: , where Indicates the final matching solution. , , are the adjustment coefficients for pain priority, patient development value, and risk control value, respectively. and Respectively represent the weighted sum of the main parameters and the weighted sum of the secondary parameters, Represents the patient's developmental value, which is used to adjust the personalized weight of the scheme matching, through the psychological development stage value and physiological development stage values The weighted sum of Represents the risk control value, which is used to constrain the management plan to ensure its safety, feasibility and patient acceptance. The calculation formula is as follows: , where is the number of constraints, represents the i-th constraint, Represents the weight corresponding to the i-th constraint.

[0083] Step S6: Real-time feedback and adjustment of pain management plan. According to the feedback data after the patient's pain management, the personalized patient sample is updated and the personalized pain management plan is adjusted in real time.

[0084] Real-time feedback and adjustment of pain management plans include:

[0085] As the treatment progresses, the real-time monitoring system continuously collects feedback data from patients. The feedback data includes pain intensity and emotional changes. Based on the feedback data, the personalized patient sample is updated in real time, and the pain assessment results are recalculated through the pain assessment model to adjust the personalized pain management plan.

[0086] Example 2, a personalized postoperative pain management system for children, see Figure 1 As shown, it includes the following modules:

[0087] A data collection module is used to collect pain assessment data, wherein the pain assessment data includes an objective assessment part and a subjective assessment part, wherein: the objective assessment part includes patient identity information, surgery information, diet and work and rest information, and medical imaging information; the subjective assessment part includes a patient part and a parent part, wherein the patient part includes the patient's preoperative anxiety information, the patient's current pain assessment information, and the patient's pain feedback imaging information, and the parent part includes the parent's preoperative anxiety information and the parent's care information; the data collection module includes but is not limited to a smart bracelet, a sensor, a facial expression recognition device, and a voice recognition device;

[0088] A developmental stage assessment module, used to conduct developmental stage assessment on patients, including a psychological assessment unit and a physiological assessment unit, which are used to assess the psychological developmental stage PS and the physiological developmental stage PH respectively. The developmental stage is determined based on age through professional assessment tools. The developmental stage assessment results are used to weight the patient portion of the pain assessment data and match the personalized pain management plan;

[0089] The sample management module includes a data preprocessing unit, a sample updating unit and a sample management unit; the data preprocessing unit is used to preprocess the pain assessment data; the sample updating unit is used to construct a personalized patient sample according to the preprocessed pain assessment data and the development stage, and to update the personalized patient sample according to the feedback data; the sample management unit is used to actively adjust the personalized patient sample;

[0090] The pain assessment module is composed of a pain assessment model, which is used to send personalized patient samples into the pain assessment model based on a hybrid deep neural network for calculation to obtain pain assessment results, which include primary parameters and secondary parameters, wherein the primary parameters include pain location, pain state, and pain degree, and the secondary parameters include pain urgency, pain level, and pain mechanism;

[0091] A program matching module, used for matching a personalized pain management program according to the pain assessment results and the developmental stage, wherein the personalized pain management program includes drug therapy and non-drug therapy, and the non-drug therapy includes physical therapy, psychological therapy, diet therapy and work and rest therapy;

[0092] The feedback adjustment module is used to update personalized patient samples and adjust personalized pain management plans in real time based on the feedback data after patient pain management.

[0093] Example 3, a method for managing postoperative pain in children using a personalized postoperative pain management system for children;

[0094] Data collection module, collection content:

[0095] Identity information: Patient B’s name (Xiao Li), gender (female), age (6 years old), hospitalization number (789123), surgical department (general surgery), and hospitalization time (3 days).

[0096] Surgical information: The type of surgery is minimally invasive surgery, the anesthesia method is general anesthesia, the operation lasts for 1 hour, and the postoperative analgesia measure is an epidural analgesia pump (drug concentration 0.1%).

[0097] Diet and daily routine information: After surgery, the three meals a day are at 7:30, 12:00, and 18:30, and the meals are low-fat liquid food (such as rice porridge, chicken soup); the daily sleep duration is 7 hours, and the deep sleep ratio is about 50%.

[0098] Medical imaging information: Postoperative CT images showed a reduction in the area of ​​inflammation and no other abnormalities.

[0099] Equipment support: The system collects patient B's activity data (about 2,000 steps per day after surgery) and sleep data (including deep sleep duration) through a smart bracelet; and collects wound healing photos and patient facial pain feedback images through a camera.

[0100] Developmental stage assessment module, psychological development stage: According to the Children's Anxiety Scale, patient B's preoperative anxiety score was 5 (mild anxiety); through the behavioral observation scale, it was assessed that his emotional response was relatively stable, his pain cognition was weak, and his psychological development age was 5.8 years old. Physiological development stage: Patient B is 6 years old, 112 cm tall, and weighs 20 kg. According to the physiological growth standard, the patient's development status is close to the average level of 6 years old, and the physiological development value K PH =0.9.

[0101] The evaluation results showed that patient B had good adaptability to his psychological state, but needed to strengthen cognitive counseling on pain.

[0102] Sample management module, the system pre-processes the pain assessment data of patient B:

[0103] Data cleaning: Remove invalid data (such as missing data due to sensor failure).

[0104] Data normalization: The patient’s dietary intake (e.g., 1200 kcal per day) and sleep duration (7 hours) are converted into feature vectors.

[0105] Weighted adjustment: According to the psychological development stage K PS =0.8 and physiological development stage K PH =0.9, weighted adjustment was performed on pain data to generate personalized patient samples.

[0106] The pain assessment module inputs patient B’s sample into the pain assessment model based on the hybrid deep neural network and obtains the following assessment results:

[0107] Main parameters: pain location was abdominal, pain status was intermittent, and pain intensity was moderate (VAS score was 6).

[0108] Secondary parameters: pain relief (S u =0.6), the pain level was superficial, and the pain mechanism was inflammation-related.

[0109] The system integrates the model results to generate a pain assessment report and notifies medical staff to review it.

[0110] The program matching module matches and adjusts personalized pain management programs based on the evaluation results generated by the system:

[0111] Drug treatment: Maintain the basal dose of the epidural analgesia pump (drug concentration 0.1%) and increase oral acetaminophen to relieve intermittent pain.

[0112] Non-pharmacological treatments: Optimizing recovery by increasing deep sleep:

[0113] Provide 20 minutes of light music in the ward every night to help sleep;

[0114] Increased frequency of psychotherapy sessions (twice daily) to help patients better express pain perception;

[0115] Adjust the diet to high protein and low fat (such as increasing the intake of milk and eggs).

[0116] Feedback adjustment module: The system monitors the patient's pain intensity (feedback as 4 via VAS score) and sleep quality (deep sleep duration increased to 4 hours) in real time through smart bracelets, cameras and other devices.

[0117] Based on the feedback data, the system re-evaluates and adjusts the plan:

[0118] Reduce the drug concentration of the epidural analgesia pump to 0.08% to gradually reduce drug dependence;

[0119] Maintain the current frequency of psychotherapy and further observe the patient's emotional response.

[0120] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes should fall within the scope of protection of the appended claims of the present invention. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. A personalized method for postoperative pain management in children, characterized in that: The following steps are involved: Step S1: Data collection, collecting pain assessment data D, the pain assessment data includes an objective assessment part and a subjective assessment part, wherein: The objective assessment part includes patient identity information, surgical information, diet and daily routine information, and medical imaging information; The subjective assessment part includes the patient part and the parent part. The patient part includes the patient's preoperative anxiety information, the patient's current pain assessment information and the patient's pain feedback imaging information. The parent part includes the parent's preoperative anxiety information and parent care information. Step S2: Developmental stage assessment, conducting a developmental stage assessment on the patient, the developmental stage includes the psychological developmental stage PS and the physiological developmental stage PH, the developmental stage is determined based on age through a professional assessment tool, and the developmental stage assessment result is used to weight the patient portion of the pain assessment data and match the personalized pain management plan; Step S3: sample construction, preprocessing the pain assessment data, and constructing a personalized patient sample based on the preprocessed pain assessment data and developmental stage; Step S4: pain assessment, sending the personalized patient sample into the pain assessment model based on the hybrid deep neural network for calculation to obtain the pain assessment result, which includes primary parameters and secondary parameters, wherein the primary parameters include pain location, pain state, and pain degree, and the secondary parameters include pain urgency, pain level, and pain mechanism; Step S5: matching a personalized pain management plan, matching a personalized pain management plan according to the pain assessment results and the developmental stage, wherein the personalized pain management plan includes drug therapy and non-drug therapy, and the non-drug therapy includes physical therapy, psychological therapy, diet therapy, and work and rest therapy; Step S6: Real-time feedback and adjustment of pain management plan: updating personalized patient samples based on the feedback data after the patient's pain management, and adjusting the personalized pain management plan in real time; The pain assessment model includes: The pain assessment model is based on a hybrid deep neural network, which is constructed by combining a convolutional neural network and a long short-term memory network. The convolutional neural network is used to extract features from the patient's medical imaging information and pain feedback imaging information; the long short-term memory network is used to process time series data and analyze the changing trend and feedback information of the patient's postoperative pain to capture the pain evolution pattern over a long time span. The structure of the pain assessment model includes: An input layer, for receiving personalized patient samples, comprising two independent input gates, for inputting data selection in the personalized patient samples into the convolution hidden layer and the memory hidden layer respectively; Hidden layer, used to extract deep features in the data, including convolution hidden layer, memory hidden layer and evaluation hidden layer. The convolution hidden layer and memory hidden layer correspond to the fully connected layer, convolution layer and pooling layer of the pre-trained convolutional neural network and long short-term memory network respectively; the evaluation hidden layer is used to comprehensively evaluate the outputs of the convolution hidden layer and the memory hidden layer; The output layer generates pain assessment parameters based on the comprehensive assessment results; The output of the pain assessment model is a multi-label classification, which is used to predict multiple pain assessment parameters, including pain location P l , pain state P s , pain level P d and pain relief u , pain level S a , pain mechanism m , among which the pain level The calculation formula is: , where and are the feature sets of the input convolutional neural network and the long short-term neural network, respectively. and are the training weights corresponding to the two models respectively. is the bias term for calculation; The loss function of the pain assessment model uses a weighted cross entropy loss function to handle the problem of multi-class imbalance. The calculation formula is: , where represents the loss function value, represents the number of categories, Represents the category index, is the weight coefficient of the i-th category, is the true label of the i-th category, is the predicted probability of the i-th category; The pain assessment model was optimized through transfer learning technology. Feature extraction was performed by pre-training convolutional neural networks and long short-term memory networks on the same sample set, and fine-tuning was performed based on the patient's personalized data to obtain a hybrid deep neural network for constructing the pain assessment model.

2. The personalized pediatric postoperative pain management method according to claim 1, characterized in that: The developmental assessment consists of two parts: Psychological development stage assessment: The patient's psychological development stage is assessed through standardized psychological assessment tools to obtain the psychological development age. The psychological assessment tools include but are not limited to the Children's Anxiety Scale, Children's Pain Coping Style Scale and Behavior Observation Scale; the results of the psychological development stage assessment include the patient's anxiety level, emotional response and cognitive pattern of pain; Physiological development stage assessment: The physiological development stage is assessed through the patient's physiological characteristics, which include age, height, weight and physiological indicators, combined with physiological development standards to determine the patient's physiological maturity; physiological assessment tools include but are not limited to age-weight curves and physiological growth standard curves.

3. The personalized pediatric postoperative pain management method according to claim 1, characterized in that: Preprocessing of pain assessment data includes: Data cleaning: remove invalid data, outliers and noise to ensure data accuracy; Data standardization: converting data from different sources into standardized feature vectors so that the data can be used as input for the pain assessment model; Weighted adjustment: The patient portion of the pain assessment data is weighted according to the child's developmental stage to ensure that the assessment results truly reflect the patient's overall condition; The weighted adjustment is calculated as: , where is the weighted and adjusted pain assessment data for item x, is the pain assessment data for item x, and They are the values ​​for the psychological development stage and the physiological development stage, respectively.

4. The personalized pediatric postoperative pain management method according to claim 1, characterized in that: Personalized pain management solutions include: Combine the pain assessment results, the patient's developmental stage assessment results, and the patient's personal preferences and medical needs to match a preliminary personalized pain management plan. The plan design includes drug therapy and non-drug therapy. According to the complexity of pain and individual differences of patients, the preliminary plan is adjusted through multi-level optimization rules, including pain priority, patient development value and risk control value. The personalized management plan is matched through the following formula: , where Indicates the final matching solution. , , are the adjustment coefficients for pain priority, patient development value, and risk control value, respectively. and Respectively represent the weighted sum of the main parameters and the weighted sum of the secondary parameters, Represents the patient's developmental value, which is used to adjust the personalized weight of the scheme matching, through the psychological development stage value and physiological development stage values The weighted sum of Represents the risk control value, which is used to constrain the management plan to ensure its safety, feasibility and patient acceptance. The calculation formula is as follows: , where is the number of constraints, represents the i-th constraint, Represents the weight corresponding to the i-th constraint.

5. The personalized pediatric postoperative pain management method according to claim 1, characterized in that: Real-time feedback and adjustment of pain management plans include: As the treatment progresses, the real-time monitoring system continuously collects feedback data from patients. The feedback data includes pain intensity and emotional changes. Based on the feedback data, the personalized patient sample is updated in real time, and the pain assessment results are recalculated through the pain assessment model to adjust the personalized pain management plan.

6. A personalized postoperative pain management system for children, characterized in that: The system applies the personalized pediatric postoperative pain management method according to any one of claims 1 to 5 above, and comprises the following modules: A data collection module is used to collect pain assessment data, wherein the pain assessment data includes an objective assessment part and a subjective assessment part, wherein: the objective assessment part includes patient identity information, surgery information, diet and work and rest information, and medical imaging information; the subjective assessment part includes a patient part and a parent part, wherein the patient part includes the patient's preoperative anxiety information, the patient's current pain assessment information, and the patient's pain feedback imaging information, and the parent part includes the parent's preoperative anxiety information and the parent's care information; the data collection module includes but is not limited to a smart bracelet, a sensor, a facial expression recognition device, and a voice recognition device; A developmental stage assessment module, used to conduct developmental stage assessment on patients, including a psychological assessment unit and a physiological assessment unit, which are used to assess the psychological developmental stage PS and the physiological developmental stage PH respectively. The developmental stage is determined based on age through professional assessment tools. The developmental stage assessment results are used to weight the patient portion of the pain assessment data and match the personalized pain management plan; The sample management module includes a data preprocessing unit, a sample updating unit and a sample management unit; the data preprocessing unit is used to preprocess the pain assessment data; the sample updating unit is used to construct a personalized patient sample according to the preprocessed pain assessment data and the development stage, and to update the personalized patient sample according to the feedback data; the sample management unit is used to actively adjust the personalized patient sample; The pain assessment module is composed of a pain assessment model, which is used to send personalized patient samples into the pain assessment model based on a hybrid deep neural network for calculation to obtain pain assessment results, which include primary parameters and secondary parameters, wherein the primary parameters include pain location, pain state, and pain degree, and the secondary parameters include pain urgency, pain level, and pain mechanism; A program matching module, used for matching a personalized pain management program according to the pain assessment results and the developmental stage, wherein the personalized pain management program includes drug therapy and non-drug therapy, and the non-drug therapy includes physical therapy, psychological therapy, diet therapy and work and rest therapy; The feedback adjustment module is used to update personalized patient samples and adjust personalized pain management plans in real time based on the feedback data after patient pain management.

Images