Intelligent monitoring system for pain assessment and management of critical patient

Abstract

The invention relates to an intelligent monitoring system for pain assessment and management of a critical patient, and relates to the technical field of pain monitoring. The assessment record extraction module extracts a pain assessment record grade and a pain source record type; the reference level statistics module counts a pain assessment reference level based on the patient portrait; the monitoring parameter calculation module determines pain monitoring personality parameters in combination with the pain assessment record level and the pain assessment reference level; the pain level prediction module predicts pain level time sequence information by taking patient portraits and physiological monitoring time sequence information as constraints; the pain level correction module corrects the pain level time sequence information according to the pain monitoring personality parameters; and the correction result is sent to the monitoring terminal through the information feedback module. The technical problem that in the prior art, a subjective and objective combined pain assessment means lacks, so that pain monitoring is insufficient in real-time performance and comprehensiveness is solved, and timely and accurate pain monitoring of the critical patient is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of pain monitoring, and in particular to an intelligent monitoring system for pain assessment and management of critically ill patients. Background Art

[0002] Pain is a common and important clinical symptom in critically ill patients. It not only affects the patient's physical health, but also has a negative impact on their mental health. Timely and accurate pain assessment and management are crucial to the patient's treatment and rehabilitation.

[0003] At present, there are two main systems for pain monitoring in critically ill patients: one is the traditional subjective scoring system, such as the numerical rating system (NRS) and the critical pain observation tool (CPOT). These systems mainly analyze the pain assessment score through the subjective description and observation of pain by patients or medical staff. Although the operation is simple, it can only be perceived when the patient's pain is more obvious, and there is a serious lag. In addition, since the subjective score depends on the patient's expression ability and the judgment of medical staff, it is easily affected by many factors, such as the patient's level of consciousness and the evaluation experience of medical staff, resulting in poor accuracy and consistency of the evaluation results. Another system is to indirectly assess the degree of pain by analyzing the patient's physiological indicators, such as heart rate, blood pressure, etc. Although this system can solve the problem of lag and improve the objectivity of pain monitoring, it cannot fully reflect the patient's pain status due to the lack of effective representation of the patient's subjective pain perception data. Summary of the invention

[0004] The present invention aims to solve the technical problem that the prior art relies only on single subjective and objective means to evaluate pain, lacks pain evaluation and management means that combine subjective and objective monitoring data, resulting in insufficient timeliness and comprehensiveness in pain evaluation of critically ill patients. An intelligent monitoring system for pain evaluation and management of critically ill patients is provided to solve the problem.

[0005] The technical solution of the present invention to solve the above technical problems is as follows: The present invention provides an intelligent monitoring system for pain assessment and management of critically ill patients, the system comprising: a portrait acquisition module, used to obtain a portrait of a critically ill patient, wherein the portrait of the critically ill patient includes gender, age, weight, pain source and pain assessment medical history; an assessment record extraction module, used to extract a number of pain assessment record levels and a number of pain source record types corresponding to each other according to the pain assessment medical history; a benchmark level statistics module, used to count a number of first pain assessment benchmark levels based on the gender, age and weight, respectively, with the several pain source record types as constraints; a monitoring parameter calculation module, used to calculate Calculate the ratio of the several pain assessment record levels to the several first pain assessment benchmark levels, and then perform mode analysis to obtain the patient's pain monitoring personality parameters; a pain level prediction module is used to predict the first pain level timing information with the gender, age, weight and pain source as background constraints and the patient's physiological monitoring timing information as foreground constraints; a pain level correction module is used to correct the first pain level timing information according to the patient's pain monitoring personality parameters to obtain second pain level timing information; an information feedback module is used to send the second pain level timing information to the patient's pain monitoring terminal.

[0006] The beneficial effects of the present invention are as follows: through the systematic operation of each module, a comprehensive assessment and management of pain in critically ill patients is achieved. The portrait acquisition module and the assessment record extraction module provide the individual characteristics and historical pain data of the patient, and the benchmark level statistics module and the monitoring parameter calculation module combine these data to establish personalized pain assessment benchmarks and parameters for each patient. The pain level prediction module and the correction module use these personalized parameters and physiological monitoring data to predict and correct the patient's pain level, thereby improving the timeliness and accuracy of pain assessment. The information feedback module feeds back the assessment results to medical staff in real time to form a closed-loop management. On the whole, this technical solution realizes the personalization, timeliness and accuracy of pain monitoring for critically ill patients, can fully reflect the patient's pain status, improves the scientificity and effectiveness of pain monitoring for critically ill patients, and helps to improve the treatment experience of critically ill patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 A schematic diagram of the structure of an intelligent monitoring system for pain assessment and management of critically ill patients provided by the present invention.

[0008] Figure 2 A schematic diagram of a process for generating a first pain assessment benchmark level in an intelligent monitoring system for pain assessment and management of critically ill patients provided by the present invention.

[0009] Figure 3A schematic diagram of a process for constructing a pain level prediction model library in an intelligent monitoring system for pain assessment and management of critically ill patients provided by the present invention.

[0010] Explanation of the reference numerals: portrait acquisition module 10 , evaluation record extraction module 20 , benchmark level statistics module 30 , monitoring parameter calculation module 40 , pain level prediction module 50 , pain level correction module 60 , information feedback module 70 . DETAILED DESCRIPTION

[0011] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0012] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0013] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in the present invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present invention.

[0014] Embodiment 1: like Figure 1 As shown, an embodiment of the present invention provides an intelligent monitoring system for pain assessment and management of critically ill patients, the system comprising: The portrait acquisition module 10 is used to obtain the portrait of the critically ill patient, wherein the portrait of the critically ill patient includes gender, age, weight, pain source and pain assessment medical history.

[0015] Specifically, the portrait of critically ill patients refers to a comprehensive description of the patient's individual characteristics and health status formed by collecting and integrating the basic information and medical data of critically ill patients, including basic information such as the patient's gender, age, weight, as well as medical data such as the source of pain and pain assessment records.

[0016] First, collect the patient's basic information such as gender, age, and weight through the medical information system or input from medical staff. For example, a 60-year-old male patient weighs 70 kg. This information can be directly obtained through the electronic medical record system. Next, extract the pain source information from the patient's medical record, such as the surgical site, trauma area, etc. For example, the patient suffered a broken leg due to a car accident, and the source of pain was the leg. At the same time, collect the patient's pain assessment medical history, including previous pain assessment records and treatment conditions. For example, the patient used the numerical rating system (NRS) to assess the pain level as 8 points when he was admitted to the hospital.

[0017] By obtaining portraits of critically ill patients, we can fully understand the patient's individual characteristics and pain history, provide basic data support for subsequent personalized pain assessment and management, and help improve the accuracy and specificity of pain assessment.

[0018] The assessment record extraction module 20 is used to extract a plurality of pain assessment record levels and a plurality of pain source record types that correspond one to one according to the pain assessment medical history.

[0019] Specifically, the pain assessment record level refers to the level of quantitative scoring of the patient's pain level based on the pain assessment tool or scale. For example, the 0-10 level assessed using the NRS. The pain source record type refers to the classification of different causes or locations that cause the patient's pain. For example, postoperative pain, traumatic pain, inflammatory pain, etc.

[0020] Extract pain scores and corresponding pain source types from the pain assessment records in the patient portrait. For unstructured text medical records, natural language processing tools (such as SpaCy or NLTK) can be used to extract keywords and match pain record levels and pain source types. For structured data, field values ​​can be directly extracted through SQL queries. The extracted data is then organized into a table for statistical analysis. These pain assessment records correspond to pain source information one-to-one, that is, each pain assessment record has a corresponding pain source information. For example, natural language processing technology is used to analyze the description in the medical record to extract information such as "the pain level on the first day after surgery was 7 points" and "the source of pain is the surgical site."

[0021] The assessment record extraction module 20 can organize the patient's pain assessment data and pain source information in a structured manner, provide accurate input data for subsequent baseline level statistics and individual parameter calculations, and help improve the efficiency and accuracy of pain assessment.

[0022] The reference level statistics module 30 is used to count a plurality of first pain assessment reference levels based on the gender, the age, and the weight, respectively, with the plurality of pain source record types as constraints.

[0023] Specifically, first, based on the patient's basic information such as gender, age, and weight, similar portraits of critically ill patients are searched from the patient database. For the portraits of critically ill patients found, the corresponding first pain assessment benchmark levels are calculated based on different pain source record types. This first pain assessment benchmark level is the average pain level obtained from the patient group and is used as a reference standard. For example, in a group of male patients over 60 years old and weighing 55kg to 60kg, the mean or median pain assessment of the group is calculated as the benchmark level through statistical analysis tools (such as Python's Pandas library or R language), and the benchmark level of postoperative pain is 6 points, and the benchmark level of traumatic pain is 5 points.

[0024] The benchmark level statistics module 30 can establish a pain assessment benchmark based on the patient's basic information, providing a comparison basis for personalized pain assessment.

[0025] The monitoring parameter calculation module 40 is used to calculate the ratio of the plurality of pain assessment record levels to the plurality of first pain assessment benchmark levels, and then perform mode analysis to obtain the patient's pain monitoring individual parameters.

[0026] Specifically, the ratio of each pain assessment record level to the corresponding first pain assessment benchmark level is calculated. All calculated ratios are subjected to mode analysis to find the ratio with the most occurrences as the patient pain monitoring personality parameter. For example, the ratios of the pain assessment record levels of a patient in multiple postoperative assessments to the corresponding first pain assessment benchmark level are 0.85, 0.86, 0.86, 0.84, and 0.86, with a mode of 0.86, indicating that the patient's pain sensitivity is 14% lower than the benchmark.

[0027] The individual parameter calculation module can quantify the differences in patients' pain perception under different pain sources, provide a personalized adjustment basis for subsequent pain level prediction and correction, and help improve the accuracy and individualization of pain assessment.

[0028] The pain level prediction module 50 is used to predict the first pain level time series information using the gender, age, weight and pain source as background constraints and the patient's physiological monitoring time series information as foreground constraints.

[0029] Specifically, the patient's physiological monitoring time series information refers to the patient's physiological monitoring data collected in chronological order, including physiological indicator parameters such as respiratory rate, blood pressure, heart rate, body temperature and oxygen content at different time points. Taking the patient's static characteristics such as gender, age, weight and pain source as the background constraints, and the patient's dynamic physiological monitoring time series information as the foreground constraints, a deep learning model is used to model the time series data, and a pain level prediction model is established to predict the changing trend of the pain level. Model training and prediction can be implemented using TensorFlow or PyTorch. For example, a long short-term memory neural network model is used to predict future pain levels based on the time series characteristics of the patient's physiological monitoring data.

[0030] The pain level prediction module 50 can comprehensively consider the individual characteristics and physiological changes of the patient, realize timely and accurate prediction of the pain level, provide forward-looking guidance for pain management, and help improve the timeliness and effectiveness of pain management.

[0031] The pain level correction module 60 is used to correct the first pain level time series information according to the patient's pain monitoring individual parameters to obtain second pain level time series information.

[0032] Specifically, the first pain level time series information refers to the initial pain level prediction result obtained in the pain level prediction process, that is, the prediction data sequence of the patient's pain level changing over time. The second pain level time series information refers to the final pain level prediction result obtained after correction based on the patient's pain monitoring personality parameters on the basis of the first pain level time series information.

[0033] The first pain level time series information output by the pain level prediction module 50, i.e., the initial pain level prediction result, is obtained, and the first pain level time series information is corrected according to the patient's pain monitoring personality parameter. For example, if a patient's pain monitoring personality parameter is 1.33, and the first pain level time series information shows that the predicted pain level at a certain moment in the future is 7 points, then the initial predicted pain level of 7 points is multiplied by 1.33, and the corrected pain level is 9.31, which is rounded to 9 points.

[0034] The pain level correction module 60 can make personalized adjustments to the pain level prediction results according to individual differences of patients, thereby further improving the accuracy and reliability of pain assessment.

[0035] The information feedback module 70 is used to send the second pain level time sequence information to the patient pain monitoring terminal.

[0036] Specifically, the patient pain monitoring terminal refers to a device or system for receiving and displaying patient pain level information. The information feedback module 70 sends the second pain level time series information, i.e., the corrected pain level prediction result, to the patient pain monitoring terminal through a network or other communication methods, such as Wi-Fi, Bluetooth, etc., to achieve rapid transmission and display of information. For example, the pain level data is transmitted to a tablet computer of a medical staff, or the changes in the patient's pain level are displayed in real time on a display screen in a ward.

[0037] The information feedback module 70 can realize real-time feedback of pain monitoring information, so that medical staff can understand the patient's pain condition in time, quickly adjust the treatment plan, and form a closed loop of pain management.

[0038] Furthermore, the benchmark level statistics module 30 is further configured to perform the following steps: Step P31: extracting a first pain source record type from the plurality of pain source record types.

[0039] Step P32: Collecting a patient pain assessment level record data set based on the gender, age, weight and the first pain source record type as constraints.

[0040] Step P33: hierarchically aggregate the patient pain assessment level record data set to generate a first pain assessment benchmark level, and add the first pain assessment benchmark levels.

[0041] Specifically, a specific pain source record type is randomly selected as the first pain source record type from among the several pain source record types obtained by the evaluation record extraction module 20. For example, the pain source record types extracted from the medical records include postoperative pain, traumatic pain, inflammatory pain, etc., and postoperative pain is selected as the first pain source record type.

[0042] Using the patient's gender, age, weight, and first pain source record type as search conditions, retrieve a pain assessment level record dataset under the same conditions from the medical information system. For example, for a 60-year-old male patient, weighing 70 kg, with postoperative pain, retrieve the pain assessment level data of all patients aged 60, male, weighing 68 to 72 kg, and with postoperative pain from the medical information database, integrate these retrieved data, and generate a patient pain assessment level record dataset. This patient pain assessment level record dataset is a data set containing many pain assessment level-related records of patients with the same gender, age, weight, and pain source record type.

[0043] For the collected data sets of patient pain assessment level records, data mining or data analysis tools (such as the Pandas library in Python) are used for hierarchical aggregation. For example, the data is first stratified according to factors such as age range and gender, and then the mean or median of the pain assessment level and other statistics are calculated in each layer. The result is used as the first pain assessment benchmark level, and then this benchmark level is added to several first pain assessment benchmark level sets that already exist or are being constructed to form a complete first pain assessment benchmark level set. By generating the first pain assessment benchmark level through hierarchical aggregation, a benchmark level can be established for each pain source under specific conditions, providing a comprehensive reference standard for subsequent pain level prediction and correction.

[0044] Further, such as Figure 2 As shown, step P33 includes: Step P331: Based on the assessment level threshold, cluster analysis is performed on the patient pain assessment level record data set, and the patient pain assessment level record data are sorted from large to small according to the number of patient pain assessment level record data in the cluster, to obtain the first cluster of patient pain assessment level record data, the second cluster of patient pain assessment level record data, and so on until the Nth cluster of patient pain assessment level record data.

[0045] Step P332: Traverse the pain assessment level record data of the second cluster of patients until the pain assessment level record data of the Nth cluster of patients, perform a primary central tendency analysis, and obtain the pain assessment benchmark level of the second cluster of patients until the pain assessment benchmark level of the Nth cluster of patients.

[0046] Step P333: Perform a secondary central tendency analysis on the pain assessment benchmark levels of the second cluster of patients up to the pain assessment benchmark levels of the Nth cluster of patients to obtain the secondary patient pain assessment benchmark levels.

[0047] Step P334: Perform a three-level central tendency analysis on the second-level patient pain assessment benchmark level and the first cluster of patient pain assessment level record data to obtain the first pain assessment benchmark level.

[0048] Specifically, the assessment level threshold refers to the critical value used to divide the pain assessment level. For example, the numerical rating scale (NRS) is used to assess the pain level, dividing the pain level into 0-3 points (mild), 4-6 points (moderate), and 7-10 points (severe).

[0049] Use a clustering algorithm (such as the K-Means clustering algorithm) to perform cluster analysis on the patient pain assessment level record data set based on the assessment level threshold, and group patients with similar pain levels into a cluster. For example, for a data set with a pain level of 0 to 10 points, cluster the patient data of 0 to 3 points into one cluster, 4 to 6 points into another cluster, 7 to 10 points into a third cluster, and so on. Then sort the patient pain assessment level record data in each cluster from large to small according to the number of patient pain assessment level record data, with the largest number of clusters in front, to obtain the first cluster of patient pain assessment level record data, the second cluster of patient pain assessment level record data, and so on until the Nth cluster of patient pain assessment level record data.

[0050] The statistical functions in the statistical analysis software (such as the mean function in the NumPy library to calculate the average value) are used to traverse the patient pain assessment level record data from the second cluster to the Nth cluster, and a primary central tendency analysis is performed on each cluster to calculate the mean or median to determine the baseline pain assessment level of the second cluster patients until the baseline pain assessment level of the Nth cluster patients. For example, the average value is calculated for the pain level data in each cluster, and this average value is used as the baseline pain assessment level of the patients in the cluster.

[0051] The pain assessment benchmark levels of the second cluster of patients are analyzed again until the pain assessment benchmark levels of the Nth cluster of patients, and the pain assessment benchmark levels are optimized to obtain the pain assessment benchmark levels of the second-level patients. For example, the average or median of the benchmark levels of these clusters is calculated again to obtain the pain assessment benchmark levels of the second-level patients.

[0052] The pain assessment benchmark level of the secondary patients and the pain assessment level record data of the first cluster of patients are subjected to a three-level central tendency analysis, and multiple benchmark levels are integrated to ensure that the final benchmark level can accurately reflect the overall data. For example, the pain assessment benchmark level of the secondary patients and the pain assessment level record data of the first cluster of patients are combined, and then the average or median and other statistics are calculated to obtain the first pain assessment benchmark level.

[0053] Through multi-step hierarchical analysis, the first pain assessment benchmark level can be more accurately derived from a complex patient pain assessment level record data set. This system fully considers the distribution characteristics of the data, avoids the deviation that may be caused by a single calculation system, and enables the benchmark level to better reflect the central trend of pain assessments of different types of patients, thereby providing more accurate basic data for subsequent pain assessments and personalized parameter calculations, helping to improve the accuracy and reliability of the entire pain assessment system.

[0054] Furthermore, the pain level prediction module 50 is further configured to perform the following steps: Step P51: Using the gender, age, weight and pain source as background constraints, index a target pain level prediction model from a pain level prediction model library.

[0055] Step P52: Processing the patient's physiological monitoring timing information through the target pain level prediction model to generate the first pain level timing information.

[0056] Specifically, the pain level prediction model library is a collection of multiple pain level prediction models, each of which is optimized and trained for different background constraints (the patient's gender, age, weight, and pain source). Based on the patient's background constraints such as gender, age, weight, and pain source, a model matching algorithm or query system is used to retrieve the model that best matches the current patient's background constraints from the pain level prediction model library as the target pain level prediction model, which is used to predict the patient's pain level.

[0057] The patient's physiological monitoring time series information is input into the target pain level prediction model. The model will predict and infer the pain level based on the mapping relationship between the patient's physiological indicator change pattern and pain perception. For example, when the patient's heart rate suddenly increases and the blood pressure fluctuates greatly, the model may predict that the patient's pain level will increase accordingly. By processing the patient's physiological monitoring time series information through the target pain level prediction model, it is possible to achieve dynamic prediction and real-time monitoring of the patient's pain level, timely capture the changing trend of pain, and provide medical staff with forward-looking pain management information, which helps to improve the timeliness and effectiveness of pain management.

[0058] Further, such as Figure 3 As shown, the pain level prediction module 50 is also used to construct a pain level prediction model library, and the execution steps include: Step P5-1: Based on the preset gender, preset age range, preset weight range and preset pain source type, collect the patient's physiological monitoring record time series information and pain level record time series information.

[0059] Step P5-2: Segment the patient physiological monitoring record timing information to obtain the front segment patient physiological monitoring record timing information and the back segment patient physiological monitoring record timing information.

[0060] Step P5-3: extracting the posterior segment pain level record timing information from the pain level record timing information according to the posterior segment patient physiological monitoring record timing information.

[0061] Step P5-4: Using the time series information of the posterior patient physiological monitoring records as supervision and the time series information of the anterior patient physiological monitoring records as input, a long short-term memory neural network is trained to obtain a primary mapping channel.

[0062] Step P5-5: Using the posterior segment pain level record timing information as supervision and the posterior segment patient physiological monitoring record timing information as input, a feedforward neural network is trained to obtain a secondary mapping channel.

[0063] Step P5-6: Merge the output layer of the primary mapping channel and the input layer of the secondary mapping channel to generate a pain level prediction model.

[0064] Step P5-7: Set the preset gender, the preset age range, the preset weight range and the preset pain source type as index tags of the pain level prediction model to construct the pain level prediction model library.

[0065] Specifically, the preset gender, preset age range, preset weight range and preset pain source type are constraints for screening patient data. With the preset gender, preset age range, preset weight range and preset pain source type as constraints, the qualified patient data is screened out from the medical database, including physiological monitoring record timing information and pain level record timing information. Among them, the patient's physiological monitoring record timing information refers to the continuous monitoring data of physiological indicators such as respiratory rate, blood pressure, heart rate, body temperature and oxygen content of the patient within a certain time range. The pain level record timing information is the continuous evaluation data of the pain level within the same time range of the corresponding patient's physiological monitoring record timing information. For example, the preset gender is male, the preset age range is 50-60 years old, the preset weight range is 60-80 kg, and the preset pain source type is postoperative pain. It is necessary to screen out the physiological monitoring record timing information and the corresponding pain level record timing information of male patients aged 50-60 years old, weighing between 60-80 kg, and suffering from postoperative pain from the hospital information system.

[0066] The collected patient physiological monitoring record timing information is segmented. It can be divided according to the number of data points or a specific time range. For example, the physiological monitoring data for 24 hours after surgery is divided into the first 12 hours and the last 12 hours. Or for an array with a data volume of N, the first N / 2 data points are used as the front-segment patient physiological monitoring record timing information, and the last N / 2 data points are used as the back-segment patient physiological monitoring record timing information. According to the time range or index of the back-segment patient physiological monitoring record timing information, the corresponding back-segment pain level record timing information is extracted from the pain level record timing information.

[0067] Use a deep learning framework (such as TensorFlow or PyTorch) to build a long short-term memory neural network. Take the time series information of the anterior patient's physiological monitoring record as input and the time series information of the posterior patient's physiological monitoring record as supervision (target output) to train the long short-term memory neural network. During the training process, the network automatically adjusts parameters such as weights to establish a mapping relationship from input to output, and finally obtains a primary mapping channel.

[0068] A feedforward neural network was constructed using a deep learning framework. The time series information of the posterior segment patient physiological monitoring records was used as input, and the time series information of the posterior segment pain level records was used as supervision. The feedforward neural network was trained so that the network could learn the mapping relationship between input and output, thereby obtaining a secondary mapping channel.

[0069] The output layer of the first-level mapping channel is merged with the input layer of the second-level mapping channel. For example, at the code level, the neurons of the two network layers are connected or a new network structure is redefined to generate a pain level prediction model.

[0070] The preset gender, preset age range, preset weight range, and preset pain source type are set as index tags of the pain level prediction model. In the construction of the model library, a database or a specific data structure can be used to store these models and their index tags. For example, a dictionary structure in Python is used, with the index tag as the key and the pain level prediction model as the value, to build a pain level prediction model library.

[0071] The pain level prediction model library constructed through the above steps can effectively train models for patients of different genders, age ranges, weight ranges, and pain source types using the patient's physiological monitoring record time series information and pain level record time series information. Each model has its specific applicable scenario (identified by an index tag), which improves the accuracy and pertinence of pain level prediction. This model library built on a neural network can better handle the complex relationships of time series data and provide a reliable tool for predicting pain levels.

[0072] Furthermore, the pain level modification module 60 is further configured to perform the following steps: Step P61: Modify the pain level threshold through the patient pain monitoring terminal configuration.

[0073] Step P62: According to the modified pain level threshold, a pain level set whose pain level is less than the modified pain level threshold is filtered from the first pain level time series information.

[0074] Step P63: Perform product calculation on the pain level set according to the patient's pain monitoring individual parameters to obtain a pain correction level set.

[0075] Step P64: Update the first pain level timing information according to the pain correction level set to obtain the second pain level timing information.

[0076] Specifically, the modified pain level threshold refers to a critical value used to determine whether the pain level needs to be modified, and the pain level below the threshold will be modified. The modified pain level threshold is input or adjusted by a medical staff or a system administrator through the user interface of the patient pain monitoring terminal.

[0077] According to the set threshold for correcting the pain level, the first pain level time series information is traversed to filter out data points with pain levels less than the threshold, thereby focusing on those pain conditions that may be underestimated, providing an accurate data basis for subsequent pain level corrections, and helping to improve the comprehensiveness and accuracy of pain assessment.

[0078] Each data point in the screened pain level set is multiplied by the patient's pain monitoring personality parameter, and the pain level is corrected according to the patient's personality parameter to obtain a pain correction level set. The data in this pain correction level set is more in line with the patient's individual pain perception, ensuring the personalization and accuracy of pain assessment.

[0079] The original data of the corresponding time point in the first pain level time series information is replaced by the data in the pain correction level set, the time series information is updated, and the second pain level time series information is obtained.

[0080] By dynamically correcting and updating in real time the data in the first pain level time series information that is less than the revised pain level threshold, we optimize the situation where the pain level is low and may be greatly affected by individual patient factors, provide more accurate and timely pain assessment information for pain management, improve the accuracy of pain level prediction, and provide medical staff with a reference basis for formulating treatment and care plans that are more in line with the patient's actual situation.

[0081] Furthermore, the information feedback module 70 is further configured to perform the following steps: Step P71: Configure the intervention pain level threshold through the patient pain monitoring terminal.

[0082] Step P72: Count the proportion of moments in the second pain level time series information when the pain level is greater than or equal to the intervention pain level threshold, and obtain an intervention recommendation degree.

[0083] Step P73: When the intervention recommendation degree is greater than or equal to the recommendation degree threshold, an intervention mark is added to the second pain level timing information and sent to the patient pain monitoring terminal.

[0084] Specifically, the intervention pain level threshold refers to the critical value at which intervention is required when the patient's pain level reaches or exceeds the threshold. The intervention pain level threshold input by the medical staff or system administrator is obtained through the user interface of the patient pain monitoring terminal. The recommendation threshold is a pre-set value used to determine whether to send an intervention mark for the second pain level timing information.

[0085] The second pain level time series information is traversed, the number of moments when the pain level is greater than or equal to the intervention pain level threshold is counted, and the proportion of the number of moments when the pain level is greater than or equal to the intervention pain level threshold to the total monitoring time is calculated as the intervention recommendation degree. The intervention recommendation degree reflects the degree of intervention required.

[0086] The calculated intervention recommendation is compared with the recommendation threshold. If the intervention recommendation is greater than or equal to the recommendation threshold, an intervention indicator is generated. This intervention indicator is used to remind medical staff that they need to intervene in the patient. It can be a specific code or mark, such as a yellow text message prompt. Then, the intervention indicator is sent to the patient pain monitoring terminal via network communication (such as Wi-Fi or Bluetooth).

[0087] By sending pain level time series information with intervention marks, medical staff can be reminded in time to pay attention to the patient's pain condition, prompting them to take appropriate intervention measures and improve the patient's treatment experience.

[0088] Furthermore, the information feedback module 70 is further configured to perform the following steps: Step P74: When the intervention recommendation degree is less than the recommendation degree threshold, the target time information whose pain level in the second pain level time series information is greater than or equal to the intervention pain level threshold is screened.

[0089] Step P75: Render the target moment information to obtain the target moment rendering identification information.

[0090] Step P76: Send the target moment rendering identification information and the second pain level timing information together to the patient pain monitoring terminal.

[0091] Specifically, when the calculated intervention recommendation is less than the recommendation threshold, it means that the patient's overall pain condition is still within the controllable range, but attention should still be paid to the pain conditions at individual moments. At this time, the specific moment information when the pain level is greater than or equal to the intervention pain level threshold is filtered out from the second pain level timing information. The filtered target moment information is rendered and marked to obtain the target moment rendering mark information. The target moment rendering mark information can be displayed using visual elements such as different colors, shapes or icons. For example, in the pain level change curve chart, the moment when the pain level is greater than or equal to the intervention pain level threshold is marked with a red circle.

[0092] The target moment information processed by the rendering mark is integrated with the second pain level timing information to form a complete pain monitoring data packet. The integrated data packet is sent to the patient pain monitoring terminal through a wireless network or other communication methods. For example, the data is transmitted to the mobile device of the medical staff through the wireless network within the hospital, and the pain level change curve with the rendering mark and the complete timing information are displayed on the device at the same time.

[0093] When the intervention recommendation level does not reach the threshold and no comprehensive intervention is required, rendering labels can enable medical staff to quickly identify the moments when the pain is more severe when viewing the pain level time series information, thereby improving the readability and comprehensibility of the information, and helping medical staff to pay attention to and deal with patients' pain problems in a timely manner.

[0094] The embodiment of the present invention provides an intelligent monitoring system for pain assessment and management of critically ill patients, which has at least the following technical effects: First, the patient portrait acquisition module 10 collects information such as the patient's gender, age, weight, pain source and pain assessment medical history, laying the foundation for subsequent personalized assessment. Then, the assessment record extraction module 20 extracts the pain assessment record level and pain source record type from the medical record. The benchmark level statistics module 30 generates multiple pain assessment benchmark levels based on the patient's basic information and pain source through hierarchical aggregation and cluster analysis systems, providing a reference standard for pain assessment. The personality parameter calculation module obtains the patient's pain monitoring personality parameters by calculating the ratio of the pain assessment record level to the benchmark level and performing mode analysis, further improving the degree of personalization of the assessment. The pain level prediction module 50 uses long short-term memory neural networks and feedforward neural networks, combined with the patient's physiological monitoring time series information and background constraints, to predict the pain level time series information, and realizes dynamic monitoring of pain changes. The pain level correction module 60 corrects and intervenes the prediction results according to the personality parameters and intervention pain level thresholds, ensuring the accuracy and timeliness of pain assessment. Finally, the information feedback module 70 feeds back the corrected pain level timing information and target moment rendering identification information to medical staff in real time, so that they can understand the patient's pain condition in time and take corresponding measures, forming a closed loop of pain management.

[0095] Overall, the embodiment of the present invention constructs a comprehensive and personalized intelligent monitoring system for pain assessment and management of critically ill patients. The system comprehensively considers the patient's basic information, pain source, pain assessment history, and physiological monitoring information, and can more timely and accurately assess and monitor the pain status of critically ill patients, improving the real-time, accuracy, and comprehensiveness of pain monitoring for critically ill patients, and providing more targeted pain assessment data for medical staff to formulate pain management plans.

[0096] Although preferred embodiments of the present invention have been described, additional changes and modifications may occur to these embodiments once those skilled in the art understand the basic inventive concepts.

[0097] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention belong to the scope of the present invention and its equivalent technologies, the present invention is also intended to include these changes and variations.

Claims

1. An intelligent monitoring system for pain assessment and management of critically ill patients, characterized in that: include: A portrait acquisition module is used to obtain a portrait of a critically ill patient, wherein the portrait of the critically ill patient includes gender, age, weight, pain source and pain assessment medical history; An assessment record extraction module, used to extract a number of pain assessment record levels and a number of pain source record types corresponding to each other according to the pain assessment medical record; A benchmark level statistics module, for counting a plurality of first pain assessment benchmark levels based on the gender, the age, and the weight, respectively, with the plurality of pain source record types as constraints; A monitoring parameter calculation module, used to calculate the ratio of the plurality of pain assessment record levels to the plurality of first pain assessment benchmark levels, and then perform mode analysis to obtain the patient's pain monitoring individual parameters; A pain level prediction module, configured to predict first pain level time series information using the gender, age, weight and pain source as background constraints and the patient's physiological monitoring time series information as foreground constraints; A pain level correction module, used for correcting the first pain level time series information according to the patient's pain monitoring individual parameters to obtain second pain level time series information; The information feedback module is used to send the second pain level timing information to the patient pain monitoring terminal.

2. The system according to claim 1, characterized in that The benchmark level statistics module is also used to perform the following steps: Extracting a first pain source record type from the plurality of pain source record types; Collecting a patient pain assessment level record data set based on the gender, age, weight and the first pain source record type as constraints; The patient pain assessment level record data set is hierarchically aggregated to generate a first pain assessment benchmark level, which is added to the plurality of first pain assessment benchmark levels.

3. The system according to claim 2, characterized in that The benchmark level statistics module is also used to perform the following steps: Based on the assessment level threshold, cluster analysis is performed on the patient pain assessment level record data set, and the patient pain assessment level record data are sorted from large to small according to the number of patient pain assessment level record data in the cluster, to obtain the first cluster of patient pain assessment level record data, the second cluster of patient pain assessment level record data, and so on until the Nth cluster of patient pain assessment level record data; Traversing the pain assessment level record data of the second cluster of patients until the pain assessment level record data of the Nth cluster of patients to perform a primary central tendency analysis, and obtaining the pain assessment benchmark level of the second cluster of patients until the pain assessment benchmark level of the Nth cluster of patients; Performing a secondary central tendency analysis on the pain assessment baseline levels of the second cluster of patients up to the pain assessment baseline levels of the Nth cluster of patients to obtain the secondary patient pain assessment baseline levels; A three-level central tendency analysis is performed on the second-level patient pain assessment benchmark level and the first cluster of patient pain assessment level record data to obtain the first pain assessment benchmark level.

4. The system according to claim 1, characterized in that The pain level prediction module is also used to perform the following steps: Using the gender, age, weight and pain source as background constraints, indexing a target pain level prediction model from a pain level prediction model library; The target pain level prediction model is used to process the patient's physiological monitoring time series information to generate the first pain level time series information.

5. The system according to claim 4, characterized in that The pain level prediction module is also used to construct a pain level prediction model library, and the execution steps include: Collect the patient's physiological monitoring record time series information and pain level record time series information based on the preset gender, preset age range, preset weight range and preset pain source type; Segmenting the patient physiological monitoring record timing information to obtain front-segment patient physiological monitoring record timing information and back-segment patient physiological monitoring record timing information; Extracting the posterior segment pain level record time sequence information from the pain level record time sequence information according to the posterior segment patient physiological monitoring record time sequence information; Taking the time series information of the posterior patient physiological monitoring record as supervision and the time series information of the anterior patient physiological monitoring record as input, a long short-term memory neural network is trained to obtain a primary mapping channel; Using the time series information of the posterior segment pain level record as supervision and the time series information of the posterior segment patient physiological monitoring record as input, training a feedforward neural network to obtain a secondary mapping channel; Combining the output layer of the primary mapping channel with the input layer of the secondary mapping channel to generate a pain level prediction model; The preset gender, the preset age range, the preset weight range and the preset pain source type are set as index tags of the pain level prediction model to construct the pain level prediction model library.

6. The system according to claim 1, wherein: The pain level modification module is also used to perform the following steps: Modifying the pain level threshold value by configuring the patient pain monitoring terminal; According to the modified pain level threshold, screening a pain level set whose pain level is less than the modified pain level threshold from the first pain level time series information; Performing product calculation on the pain level set according to the patient pain monitoring individual parameters to obtain a pain correction level set; The first pain level timing information is updated according to the pain modification level set to obtain second pain level timing information.

7. The system according to claim 1, characterized in that The information feedback module is also used to perform the following steps: configuring intervention pain level thresholds through the patient pain monitoring terminal; Counting the proportion of moments in which the pain level in the second pain level time series information is greater than or equal to the intervention pain level threshold, and obtaining an intervention recommendation degree; When the intervention recommendation degree is greater than or equal to the recommendation degree threshold, an intervention mark is added to the second pain level timing information and sent to the patient pain monitoring terminal.

8. The system according to claim 7, characterized in that The information feedback module is also used to perform the following steps: When the intervention recommendation degree is less than the recommendation degree threshold, screening the target time information in the second pain level time series information where the pain level is greater than or equal to the intervention pain level threshold; Rendering the target time information to obtain target time rendering identification information; The target moment rendering identification information and the second pain level timing information are sent together to the patient pain monitoring terminal.

Images