Platform and method for whole-process patient self-control analgesia management based on artificial intelligence

Abstract

The invention discloses a whole-process patient self-control analgesia management platform and method based on artificial intelligence, and relates to the related field of analgesia management, and the platform comprises a multi-source data interaction module which is used for building a time sequence standard data set; the information configuration module is used for starting the self-control analgesia pump; the AI pain evaluation module is used for generating a pain scoring result; and the identification early warning module is used for generating graded early warning. The technical problems of insufficient safety and unstable effect in the analgesia management process due to dependence on a fixed dose threshold and lagged artificial pain assessment in traditional patient self-control analgesia management are solved, the whole-process pain management platform based on artificial intelligence is constructed, multi-source data can be analyzed in real time, and the pain management efficiency is improved. And the pain state of the patient is dynamically evaluated, and the parameters of the analgesia pump are automatically adjusted, so that the technical effects of personalized analgesia management and safety and effectiveness in the analgesia process are realized.

Description

Technical Field

[0001] This application relates to fields related to analgesia management, and specifically to a patient-controlled analgesia management platform and method based on artificial intelligence throughout the entire process. Background Art

[0002] In the current field of medical pain management, while patient-controlled analgesia (PCA) technology provides patients with a degree of self-directed pain control, it still presents numerous clinical challenges that need to be addressed. Traditional PCA systems primarily rely on pre-set fixed dosage parameters and subjective patient feedback. This mechanical approach to pain management has significant limitations in practical applications. Healthcare professionals often struggle to obtain timely information about patients' real-time pain status, resulting in significant lags in pain assessment. Patients often experience unnecessary pain due to a lack of timely analgesia, or experience adverse reactions such as overanalgesia due to untimely dosage adjustments. Furthermore, existing PCA systems lack the ability to intelligently analyze patients' physiological states, making it impossible to dynamically adjust safety thresholds based on individual patient differences. Furthermore, they are unable to integrate and analyze multi-source data, creating safety risks in the analgesia management process. In clinical practice, healthcare professionals frequently need to manually record and adjust their data, increasing their workload and hindering the implementation of precise, personalized analgesia management. Furthermore, decentralized medical equipment and isolated data systems hinder effective information integration, severely limiting the efficiency and quality of analgesia management. With the in-depth application of artificial intelligence technology and Internet of Things technology in the medical field, the development of an analgesia management platform that can achieve full-process, intelligent, and precise control has become an urgent need to improve the safety and effectiveness of patient analgesia. Summary of the Invention

[0003] This application provides an artificial intelligence-based full-process patient-controlled analgesia management platform and method to solve the technical problems of insufficient safety and unstable effects in the analgesia management process due to reliance on fixed dose thresholds and delayed manual pain assessment in traditional patient-controlled analgesia management. The application achieves the goal of building an artificial intelligence-based full-process pain management platform that can analyze multi-source data in real time, dynamically evaluate the patient's pain status, and automatically adjust the analgesia pump parameters, thereby realizing personalized analgesia management and ensuring the safety and effectiveness of the analgesia process.

[0004] The present application provides a patient self-controlled analgesia management platform based on artificial intelligence throughout the entire process, including: a multi-source data interaction module for accessing multi-source devices, reading device data from multi-source devices, and establishing a time-series standard data set, wherein the multi-source devices include monitors, self-controlled analgesia pumps, and wearable devices; an information configuration module for configuring the operating parameters of the self-controlled analgesia pump and then starting the self-controlled analgesia pump by an operator with preset permissions; an AI pain evaluation module for parsing the updated time-series standard data set, extracting physiological data sets, facial expression data sets, and voice data sets, and calling the patient's basic information and self-evaluation data, activating a multi-dimensional pain evaluation layer to perform pain scoring based on the physiological data set, facial expression data set, voice data set, the patient's basic information, and self-evaluation data, and generating a pain scoring result; an identification and early warning module for receiving the pain scoring result and the patient's self-controlled analgesia pump adjustment request, configuring a safety threshold based on the patient's basic information, and generating a graded early warning using the pain scoring result, the self-controlled analgesia pump adjustment request, and the safety threshold.

[0005] The present application also provides a method for patient self-controlled analgesia management based on artificial intelligence throughout the entire process, including: accessing multi-source devices, reading device data from the multi-source devices, and establishing a time-series standard data set, wherein the multi-source devices include a monitor, a self-controlled analgesia pump, and a wearable device; after configuring the operating parameters of the self-controlled analgesia pump, an operator with preset permissions starts the self-controlled analgesia pump; parsing the updated time-series standard data set, extracting the physiological data set, facial expression data set, and voice data set, and calling the patient's basic information and self-assessment data, activating the multi-dimensional pain assessment layer to perform pain scoring based on the physiological data set, facial expression data set, voice data set, the patient's basic information and self-assessment data, and generating a pain scoring result; receiving the pain scoring result and the patient's self-controlled analgesia pump adjustment request, and configuring a safety threshold based on the patient's basic information, and using the pain scoring result, the self-controlled analgesia pump adjustment request, and the safety threshold to generate a graded warning.

[0006] The present application proposes a patient self-controlled analgesia management platform and method based on artificial intelligence throughout the entire process. The multi-source devices are accessed through the multi-source data interaction module, the device data of the multi-source devices are read, and a time series standard data set is established. The multi-source devices include monitors, self-controlled analgesia pumps, and wearable devices. After the operating parameters of the self-controlled analgesia pump are configured through the information configuration module, the self-controlled analgesia pump is started by an operator with preset permissions. The updated time series standard data set is parsed through the AI pain evaluation module, the physiological data set, facial expression data set, and voice data set are extracted, and the patient's basic information and self-evaluation data are called to activate the multi-dimensional pain evaluation layer based on the physiological data set. , facial expression dataset, voice dataset, basic information of the patient and pain score of self-assessment data to generate a pain score result; receive the pain score result and the patient's automatic analgesia pump adjustment request through the identification and early warning module, and configure the safety threshold according to the patient's basic information, and use the pain score result, the automatic analgesia pump adjustment request and the safety threshold to generate a graded early warning, so as to build a full-process pain management platform based on artificial intelligence, which can analyze multi-source data in real time, dynamically evaluate the patient's pain status, and automatically adjust the analgesia pump parameters, thereby realizing personalized analgesia management and ensuring the safety and effectiveness of the analgesia process. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following is a brief introduction to the drawings of the embodiments of the present invention. Flowcharts are used in this application to illustrate the operations performed by the platform according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in exact sequence. On the contrary, various steps may be processed in reverse order or simultaneously as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0008] Figure 1 A schematic structural diagram of a patient-controlled analgesia management platform based on artificial intelligence throughout the entire process is provided in an embodiment of the present application.

[0009] Figure 2 A flowchart of a patient-controlled analgesia management method based on artificial intelligence throughout the entire process is provided in an embodiment of the present application.

[0010] Explanation of the accompanying symbols: multi-source data interaction module 11, information configuration module 12, AI pain evaluation module 13, identification and warning module 14. DETAILED DESCRIPTION

[0011] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0012] In order to make the purpose, technical solutions and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0013] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms “first\second” involved are merely to distinguish similar objects and do not represent a specific ordering of the objects. The terms “including” and “having” and any variations are intended to cover non-exclusive inclusions, for example, a process, method, platform, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. The terms used herein are for the purpose of describing the embodiments of this application only.

[0014] The present application embodiment provides a patient-controlled analgesia management platform based on artificial intelligence throughout the entire process, such as Figure 1 As shown, the platform includes:

[0015] The multi-source data interaction module 11 is used to access multi-source devices, read device data of the multi-source devices, and establish a time series standard data set. The multi-source devices include monitors, patient-controlled analgesia pumps, and wearable devices.

[0016] Specifically, the multi-source data interaction module 11 is a core component of the patient-controlled analgesia management platform, responsible for data exchange and collaboration with various devices. This module connects to a variety of medical devices through standardized communication protocols (such as HL7 and Bluetooth). These include monitors for real-time vital sign monitoring, patient-controlled analgesia pumps (PCAs) for controlling drug infusions, wearable devices (such as smart bracelets) for tracking patient activity, and bedside integrated terminals (including cameras and microphones). These devices continuously monitor patients' vital signs, medication usage records, and pain feedback data, providing a foundation for subsequent intelligent analysis. Specifically, monitors are responsible for providing physiological parameters such as heart rate, blood pressure, and blood oxygen saturation; PCAs record medication usage and device status information; wearable devices further provide data such as activity status and temperature changes; and bedside integrated terminals are used to capture patient voice and facial expressions. These data from different devices are aggregated through the multi-source data interaction module 11 to form a complete time-series standard data set, which provides important input information for subsequent AI pain assessment and intelligent control. Through standardized data interfaces, the module can ensure that data from various devices can be transmitted and integrated in a timely and accurate manner to support the entire process of analgesia management.

[0017] The information configuration module 12 is used to configure the operating parameters of the automatic analgesia pump and then start the automatic analgesia pump by an operator with preset authority.

[0018] Specifically, the information configuration module 12 is the core control unit of the platform, mainly used to set and adjust the operating parameters of the patient-controlled analgesia pump (PCA) and ensure strict management of operating permissions. The module adopts a hierarchical permission design, which subdivides operating permissions into three levels: patients, nurses, and doctors. Patients can only view their own analgesia status and trigger on-demand medication; nurses can set routine parameters such as the basal infusion rate and single-dose dosage; doctors have the highest authority and can modify safety thresholds (such as the maximum hourly dose limit), review AI score deviations, and authorize emergency operations. Through this module, operators can set parameters such as the drug composition, dosage, and frequency of use of the PCA pump according to the patient's specific situation and needs. After these parameters are configured, only operators with preset permissions can start the PCA pump, ensuring the safety and compliance of device operation. The design of this module not only ensures that the patient's personalized analgesia needs are met, but also ensures that only authorized medical personnel can adjust and start the device through permission management, avoiding potential risks caused by unauthorized operation. This function ensures the automation and accuracy of the entire analgesia management process, providing patients with a safe and personalized analgesia experience.

[0019] The AI pain evaluation module 13 is used to parse the updated time-series standard data set, extract the physiological data set, facial expression data set, and voice data set, and call the patient's basic information and self-assessment data, activate the multidimensional pain evaluation layer to perform pain scoring based on the physiological data set, facial expression data set, voice data set, the patient's basic information and self-assessment data, and generate a pain scoring result.

[0020] Specifically, the function of the AI pain evaluation module 13 is to parse the updated time-series standard data set and extract various types of data therein, including physiological data sets, facial expression data sets, and voice data sets. At the same time, the module will also combine the patient's basic information and self-assessment data to activate the multidimensional pain assessment layer, and conduct a comprehensive analysis and evaluation of the patient's pain condition from multiple dimensions. Specifically, the module comprehensively understands the patient's pain condition by parsing physiological data sets (such as heart rate, blood pressure, respiratory rate, etc.), facial expression data sets (such as pain-related facial muscle changes), and voice data sets (such as emotional analysis of tone and keywords). At the same time, the patient's self-assessment (such as the patient's pain score VAS / NRS) and basic information (such as age, medical history, etc.) are also taken into consideration to improve the accuracy of the assessment. Subsequently, the pain facial expression recognition channel in the multidimensional pain assessment layer is used based on the FACS (Facial Action Coding System) algorithm to analyze the patient's facial micro-expressions through real-time video streams, focusing on monitoring pain characteristic action units such as frowning (AU4) and eyelid closure (AU6 / 7); the physiological recognition channel is used based on machine learning models (such as LSTM neural networks) to continuously monitor the patient's vital signs such as heart rate variability and respiratory rate, and identify and capture pain-related physiological characteristic changes; the language sentiment analysis channel is used through natural language processing technology to extract acoustic features such as keyword frequency and tone fluctuations from the patient's voice; the comprehensive analysis channel is used to integrate through an adaptive weighted fusion algorithm, and finally automatically calculate a pain score result between 0 and 10 points. This result will provide doctors with real-time pain status feedback, thereby helping doctors make more accurate analgesia adjustments.

[0021] In one possible implementation, the multidimensional pain assessment layer includes:

[0022] The pain facial expression recognition channel 131 is used to receive the facial expression data set, sort the facial expression data in chronological order, capture continuous changes in facial movements based on the sorting result, establish a first recognition result based on the continuous changes in facial movements captured result, obtain the patient's standard facial expression, compare the facial expression data set with the standard facial expression, establish a second recognition result, and feed back the first recognition result and the second recognition result to the storage center of the multidimensional pain assessment layer.

[0023] Specifically, in the multidimensional pain assessment layer of the AI pain evaluation module 13, the pain facial expression recognition channel 131 receives a facial expression data set: the channel first receives the patient's facial expression data set from the multi-source data interaction module 11. This data set consists of facial expression information captured by a monitoring device (such as a camera), including changes and movements of the patient's facial muscles. The received facial expression data is then sorted in chronological order to ensure that the facial expression data can reflect the continuous changes in the patient's facial expression during analysis and facilitate subsequent processing. Subsequently, based on the sorted facial expression data, the short-term bursts and continuous changes of facial movements are analyzed according to the set first recognition constraints and second recognition constraints. This process aims to analyze the dynamic changes of facial expressions within a certain period of time, such as the raising of eyebrows, changes in the corners of the mouth, etc., thereby generating a pain probability score, and using the score as the first recognition result to represent the patient's pain expression. Afterwards, in a pain-free state (such as when the patient is calm), their standard facial expressions (such as naturally relaxed facial features) are collected as a comparison benchmark, and then the facial expression dataset is analyzed for differences with the baseline expression. For example, the contraction amplitude of AU4 in pain is calculated differentially from the baseline state, and then the calculated difference is calculated relative to the baseline state to determine the deviation of each action unit (such as AU4, AU6 / 7). By weighting these deviations, the expression deviation is determined and used as the second recognition result to reflect the matching degree of the patient's facial expression with the standard facial expression and the degree of pain. Finally, the first recognition result and the second recognition result will be fed back to the storage center of the multidimensional pain assessment layer for subsequent comprehensive analysis. These results will be used together with other data (such as physiological data and voice data) to help the AI system conduct a comprehensive and accurate assessment of the patient's pain.

[0024] In a possible implementation, in the pain facial expression recognition channel 131, establishing the first recognition result based on the result of capturing the continuous change of facial movements further includes:

[0025] Establish a first recognition constraint and a second recognition constraint, wherein the first recognition constraint is a recognition constraint for short bursts, and the second recognition constraint is a recognition constraint for continuous changes; use the first recognition constraint and the second recognition constraint to perform trigger analysis on the capture results of continuous changes in facial movements, and perform weighted analysis of short bursts and continuous changes based on the trigger analysis results to establish the first recognition result.

[0026] Specifically, in the pain facial expression recognition channel 131, in order to distinguish short-term bursts from continuous changes in facial movements from facial expression data, it is necessary to establish a first recognition constraint and a second recognition constraint, where a short-term burst refers to a sudden and rapid movement in a facial expression, such as a sudden rise of eyebrows or a momentary narrowing of eyes, which is usually due to sudden emotional reactions such as severe pain or surprise; the first recognition constraint is to capture and analyze these facial expressions with high intensity changes in a short period of time, including an instantaneous intensity threshold and a time window, such as AU4 intensity rising from level 1 to level 4 within 1 second; continuous changes refer to changes in facial expressions occurring over a longer time range, usually reflecting chronic or continuous pain, for example, facial tension, clenched lips, or a gradual increase in facial wrinkles. These changes usually last for a long time and have a relatively small amplitude; the second recognition constraint is used to capture this continuous and gradual change in facial movements, including a duration threshold and an average intensity threshold, such as AU6 / 7 continuous activation for more than 10 seconds and an average intensity ≥ level 3. Subsequently, the intensity change rate of the action unit in a short time window (such as 1 second) is calculated. If the change rate exceeds the first recognition constraint, it is determined to be a short-term burst action, and the current intensity change amplitude is recorded as the short-term burst score. Similarly, the cumulative activation time and average intensity of the action unit in a long time window (such as 15 seconds) are counted and compared with the second recognition constraint to determine the continuously changing action. The ratio of the current cumulative activation time to the continuous time threshold is calculated, and the ratio of the current average intensity to the average intensity threshold is calculated. The two ratios are added together and recorded as the continuously changing score. Afterwards, the short-term burst score and the continuously changing score are weighted and the first recognition result is established based on the weighted result. This result is a preliminary assessment of the pain intensity in the patient's facial expression, which can reflect the patient's violent reaction in a short period of time and long-term expression changes, thereby helping to accurately assess the patient's pain level and provide more accurate data support for pain assessment.

[0027] In one possible implementation, the multidimensional pain assessment layer further includes:

[0028] The physiological recognition channel 132 is used to perform heart rate variability and respiratory rate analysis on the physiological data set, generate a third recognition result, and feed the third recognition result back to the storage center of the multidimensional pain assessment layer.

[0029] Specifically, in the physiological recognition channel 132 of the multidimensional pain assessment layer, a physiological data set is extracted from the multi-source data interaction module 11, including physiological data such as heart rate variability, respiratory rate, and blood oxygen saturation. By inputting the physiological data set into the LSTM model built into the physiological recognition channel 132, the LSTM model performs pain analysis on these physiological data based on the learned mapping relationship, determines the patient's pain physiological index, and feeds back the pain physiological index as the third recognition result to the storage center of the multidimensional pain assessment layer for fusion with other pain data, thereby effectively overcoming the limitation that a single signal is susceptible to interference.

[0030] The LSTM model, consisting of an input layer, an LSTM layer, and an output layer, is trained by inputting historical physiological data and historical pain indices into the model. The model undergoes iterative training, including forward propagation, loss calculation (mean squared error), backpropagation, and parameter optimization (Adam), to achieve more accurate predictions. After training, the model is evaluated using untrained data. The loss function is calculated to determine the model's performance, and adjustments are made as needed, such as adjusting the LSTM model's hyperparameters (such as the learning rate, number of LSTM units, and hidden layer size).

[0031] In one possible implementation, the multidimensional pain assessment layer further includes:

[0032] The language sentiment analysis channel 133 is used to read the voice data set, establish a fourth recognition result based on language keywords and intonation analysis, and feed the fourth recognition result back to the storage center of the multidimensional pain assessment layer; the comprehensive analysis channel 134 is used to read the first recognition result, the second recognition result, the third recognition result, and the fourth recognition result stored in the storage center, perform joint analysis, and establish a pain score result.

[0033] Specifically, in the language sentiment analysis channel 133 of the multidimensional pain assessment layer, a patient's speech dataset is received from the multi-source data interaction module 11. This speech dataset contains the patient's speech content, including emotional expression, tone changes, speaking speed, pitch, and other information. Subsequently, the speech dataset is preprocessed, including noise removal, speech segmentation, and audio feature extraction. Common audio feature extraction methods include Mel-Frequency Cepstral Coefficients (MFCCs), pitch, and volume. After preprocessing, natural language processing (NLP) technology is used to extract key emotional vocabulary and sentences from the patient's speech. Specific emotional keywords (such as pain, discomfort, and no, etc.) are often associated with pain perception, and the frequency, pitch, and volume of the keywords are calculated. Next, a sentiment dictionary (such as VADER or SentiWordNet) is used to obtain the weight of each keyword. The weight of each keyword is multiplied by the corresponding frequency of occurrence to obtain the keyword sentiment score. The pitch and volume are then mapped to their respective score tables to obtain the pitch score and volume score. By weighting the keyword emotion score, tone score, and volume score, a pain emotion score is obtained and fed back as the fourth recognition result to the fourth recognition result. This fourth recognition result reflects the patient's emotional intensity and indication of pain perception in the voice. Then, in the comprehensive analysis channel 134 of the multidimensional pain assessment layer, the four categories of recognition results are read from the storage center and mapped to 0 to 10 for easy comprehensive calculation. The weights recorded in the comprehensive analysis channel 134 are then combined with the four categories of recognition results to perform a joint analysis, that is, the final pain score result is calculated by weighted fusion. This pain score result is a numerical value that represents the patient's current pain intensity, which may be in the range of 0 to 10 (0 means no pain, 10 means extreme pain). For example, if both facial expression and voice analysis show that the patient is in acute pain, the heart rate variability is low and the respiratory rate is fast, a higher pain score (such as 8 / 10) may be obtained. In summary, through the above-mentioned multi-channel data complementarity, the problem of single modality being susceptible to interference is effectively solved, ensuring the safety and effectiveness of the analgesia process.

[0034] The identification and warning module 14 is used to receive the pain score result and the patient's patient-controlled analgesia pump adjustment request, configure a safety threshold based on the patient's basic information, and generate a graded warning using the pain score result, the patient-controlled analgesia pump adjustment request, and the safety threshold.

[0035] Specifically, the recognition and warning module 14 first receives the pain score generated by the multidimensional pain assessment layer, and then receives the patient's request for adjustment of the patient-controlled analgesia pump, that is, the patient's analgesia demand triggered by the mobile terminal (such as clicking the additional dose button). Subsequently, the safety threshold is configured based on the patient's basic information (such as age, weight, medical history, etc.). This information helps to set the appropriate range of drug use. For example, for lighter patients, the dose of each medication will be reduced; for elderly patients or those with respiratory diseases, more stringent safety thresholds will be set based on the drug's metabolism. The safety thresholds include a maximum single dose (bolus) and a minimum safety interval (lockout interval). These parameters are used to ensure that the patient does not overuse the drug within the specified time. When performing multi-level warnings, the recognition and warning module 14 will determine the relationship between the pain score and the drug dose requested by the patient. If the patient's pain score is high (such as 7 / 10 or above) and the drug requests are frequent, it means that the patient may not be receiving adequate analgesia. At this time, the system will determine whether the analgesic dose needs to be increased or the regimen needs to be adjusted. Conversely, if the pain score is low and the patient frequently requests medication, the system alerts the patient to the risk of over-dependence and recommends that medical staff reassess their pain management plan. The requested medication dose is then compared with the safety threshold. If the patient's requested dose exceeds the preset maximum single dose, a yellow alert is triggered, reminding medical staff to review medication usage. If the medication request exceeds the minimum safety interval, a red alert is triggered, warning of the risk of overdose and automatically locking the patient-controlled analgesia pump to prevent further medication. Finally, the generated alert information is notified to medical staff via mobile devices or workstations, providing real-time feedback. A yellow alert reminds medical staff to pay attention to the patient's medication needs and check whether they are approaching the safety threshold; a red alert immediately locks the analgesia pump and issues an emergency notification requesting medical staff intervention. This hierarchical alert mechanism ensures the safety of analgesia management.

[0036] In a possible implementation, the identification and warning module 14 includes:

[0037] The information extraction submodule 141 is used to extract age information, weight information, and medical history information from the basic information; the adaptive threshold setting module 142 is used to perform threshold matching based on the age information, weight information, and medical history information, and perform threshold backoff and retention on the threshold matching results to establish the safety threshold.

[0038] Specifically, the main task of the information extraction submodule 141 is to extract key information from the patient's basic information, including age, weight, and medical history. This information is crucial for setting personalized safety thresholds. Age information refers to the patient's age data, which has a great impact on drug metabolism, dosage requirements, and side effects. Weight information refers to the patient's weight data, which is usually directly related to the calculation of drug dosage. Patients with larger or smaller weights may need to adjust the drug dosage. Medical history information refers to the patient's historical condition, including whether there are chronic diseases (such as diabetes, heart disease, respiratory diseases, etc.) or special medical history (such as allergies, drug dependence, etc.). This information guides drug selection, dosage, and the prevention of potential side effects. Subsequently, the adaptive threshold setting module 142 dynamically adjusts the drug safety threshold based on the extracted patient basic information (age, weight, medical history, etc.). Specifically, the basic range of drug dosage is determined according to the patient's age. For patients of different age groups, the safe dosage range of drugs is usually different. For example, elderly patients may need to reduce drug dosage due to slower metabolism; while the drug dosage of pediatric patients often needs to be adjusted based on the weight and age ratio. The established safety thresholds are then adjusted based on weight. Generally speaking, heavier individuals may require higher drug doses, while lighter individuals may require reduced doses. For example, the maximum bolus dose threshold is adjusted based on the weight deviation and, if not, by multiplying the deviation by the bolus dose unit (e.g., 0.02 mg). Additional safety thresholds are then set based on the patient's medical history for specific medical conditions or drug allergies. For example, patients with heart disease or respiratory conditions may require reduced doses of certain medications to avoid side effects. For patients with a history of drug allergies, medications that may trigger allergic reactions are excluded. After threshold matching is complete, the calculated threshold matching results are backed off. This process ensures that overly aggressive or unsafe drug doses are not set. This backing off helps mitigate the risk of overdosing in edge cases. Ultimately, patient-specific safety thresholds are generated based on these adjusted thresholds (including maximum bolus dose and minimum safety interval). These thresholds are used to control the operating parameters of the patient-controlled analgesia pump (PCA pump), ensuring safe and effective analgesia management tailored to the patient's specific needs and avoiding the risk of overdosing.

[0039] In a possible implementation, the identification and warning module 14 further includes:

[0040] The comparison submodule 143 is used to perform a threshold trigger analysis on the adjustment request of the self-controlled analgesia pump according to the safety threshold, and generate a first warning according to the threshold trigger analysis result; the pain warning submodule 144 is used to perform pain warning identification according to the pain scoring result, establish a second warning, and generate a graded warning according to the first warning and the second warning.

[0041] Specifically, the comparison submodule 143 receives the patient's automatic analgesia pump adjustment request and the patient's safety threshold. The automatic analgesia pump adjustment request usually includes information such as the patient's requested drug dosage and dosing frequency, while the safety threshold is a personalized drug safety range set based on the patient's basic information (such as age, weight, medical history, etc.). After receiving this information, the comparison submodule performs a threshold trigger analysis on the patient's requested drug dosage and the safety threshold. If the patient's drug request exceeds the set maximum single dose or the requested drug use time interval is less than the set minimum safety interval, it indicates that the patient's drug request has a potential risk. At this time, a first warning will be generated, which is divided into yellow warning and red warning. If the drug request is close to the safety threshold but has not yet exceeded it, a yellow warning will be generated to remind medical staff to pay attention to drug use; if the drug request exceeds the safety range, a red warning will be generated, automatically locking the analgesia pump and notifying medical staff, indicating that there may be a drug overdose or other risks. Subsequently, the pain warning submodule 144 performs further analysis based on the pain score results. The pain score reflects the patient's current pain intensity. If the pain score is high (such as more than 7 / 10) and the patient frequently requests medication, it will be identified as severe pain and insufficient analgesic effect, and a second warning (red) will be generated. If the pain score is low but the patient still frequently requests medication, it may indicate that the patient is dependent on medication, and a second warning (yellow) will be generated, indicating excessive medication use. Afterwards, based on the first warning and the second warning, the final graded warning is generated. If the first warning and the second warning are both yellow, the graded warning is yellow. If either the first warning or the second warning is red, the graded warning is red. Finally, all generated warning information will be pushed to medical staff to help them understand the patient's pain condition and medication usage status in real time, ensuring the safety and effectiveness of the patient's analgesia management.

[0042] In a possible implementation, the adaptive threshold setting module 142 is further configured to:

[0043] A pain coefficient is established according to the pain scoring result; a safety range base value is configured based on the threshold matching result, and a safety threshold is reconstructed through the pain coefficient and the safety range base value.

[0044] Specifically, the ratio of the pain score result to the maximum pain score result is calculated to establish a pain coefficient. The pain coefficient is a numerical value that measures the intensity of pain and can accurately reflect the patient's pain intensity. Subsequently, the basic value of the safety range is configured according to the threshold matching result obtained from the adaptive threshold setting module. The threshold matching result mainly considers the patient's basic information (such as age, weight, medical history, etc.), and calculates the maximum safe dose, minimum safe interval and other safety parameters of the drug based on this information. On this basis, the basic value of the safety range is generated through these configured safety range parameters (for example, the maximum single dose and the minimum safe interval). This basic value is a preliminary safety parameter for the current patient to ensure that the patient will not be at risk due to overdose or over-reliance on drugs during the entire analgesia management process. After obtaining the pain coefficient and safety range baseline values, the pain coefficient is used to adjust or reconstruct the originally set safety threshold to ensure that while providing effective analgesia, the risk of drug overdose or side effects is avoided. Specifically, the safety range baseline value is adjusted based on the pain coefficient. For example, if the patient's pain coefficient is higher than the pain neglect limit, the maximum single dose in the safety range baseline value is multiplied by the pain coefficient, and the product is then added to the maximum single dose to obtain the reconstructed maximum single dose. Similarly, the minimum safety interval in the safety range baseline value is multiplied by the pain coefficient, and the difference between the minimum safety interval and the product is calculated to obtain the reconstructed minimum safety interval. Finally, the reconstructed maximum single dose and minimum safety interval are integrated to complete the reconstruction of the safety threshold. This reconstructed safety threshold will be applied in real time during the patient's analgesia process and dynamically adjusted as the pain intensity changes, ensuring that the patient's medication use always remains within the safe range.

[0045] In a possible implementation, the platform further includes:

[0046] The device collaboration module 145 is used to generate a collaborative stop instruction when it detects that the monitored respiratory rate of the monitors of the multi-source devices meets the preset frequency threshold, use the collaborative stop instruction to control the automatic analgesia pump to pause, and issue an early warning.

[0047] Specifically, the task of the device collaboration module 145 is to monitor the physiological data provided by multiple source devices (such as monitors), especially the respiratory rate. Once it is detected that the respiratory rate exceeds the preset safety threshold, it will immediately trigger a collaborative stop instruction and control the automatic analgesia pump to pause, so as to avoid adverse reactions caused by drug overdose, and at the same time send an early warning notification to medical staff. Specifically, the device collaboration module receives real-time data from the monitor, especially respiratory rate data. The monitor continuously monitors the patient's vital signs and transmits its data to the device collaboration module. Subsequently, the device collaboration module compares the received respiratory rate with the preset frequency threshold. The threshold is usually divided into three levels: normal range, warning range and danger range. The respiratory rate within the safe range is usually 12 to 20 times / minute. A rate below this range may be a signal of respiratory depression. Exceeding a certain high threshold may mean that the patient is in acute pain or other stress reactions. When it is detected that the patient's respiratory rate exceeds the set safety range, the device collaboration module will automatically generate a collaborative stop instruction to suspend the drug administration operation of the automatic analgesia pump to prevent serious side effects caused by drug overdose. After generating a collaborative stop command, the device collaborative module will interact with the automated analgesia pump through the control interface to suspend medication. The automated analgesia pump will stop using the current medication to ensure that the patient no longer receives medication when the respiratory rate is abnormal. At the same time, the device collaborative module will also generate an early warning notification and push the early warning information to medical staff in real time. The early warning information usually includes the patient's current respiratory rate, medication status, and potential health risks. For example, if the patient's current respiratory rate is 7 times / minute, which is lower than the set safety threshold, the automated analgesia pump has been suspended. Through the early warning notification, medical staff can respond in a timely manner and take necessary measures, such as checking the patient's vital signs and adjusting the analgesia management plan, thereby improving patient safety and optimizing analgesia management.

[0048] In the above, refer to Figure 1 A patient-controlled analgesia management platform based on artificial intelligence throughout the entire process according to an embodiment of the present invention is described in detail. Figure 2 A method for patient-controlled analgesia management based on artificial intelligence throughout the entire process according to an embodiment of the present invention is described.

[0049] According to an embodiment of the present invention, an artificial intelligence-based full-process patient-controlled analgesia management method is used to solve the technical problems of insufficient safety and unstable effects in the analgesia management process due to reliance on fixed dose thresholds and delayed manual pain assessment in traditional patient-controlled analgesia management, so as to achieve the technical effect of building an artificial intelligence-based full-process pain management platform that can analyze multi-source data in real time, dynamically evaluate the patient's pain status, and automatically adjust the analgesia pump parameters, thereby realizing personalized analgesia management and ensuring the safety and effectiveness of the analgesia process.

[0050] A patient-controlled analgesia management method based on artificial intelligence throughout the entire process includes:

[0051] S100: Access multi-source devices, read device data of multi-source devices, and establish a time-series standard data set, wherein the multi-source devices include a monitor, a self-controlled analgesia pump, and a wearable device; S200: After configuring the operating parameters of the self-controlled analgesia pump, the self-controlled analgesia pump is started by an operator with preset permissions; S300: Parse the updated time-series standard data set, extract the physiological data set, facial expression data set, and voice data set, and call the patient's basic information and self-assessment data, activate the multi-dimensional pain assessment layer to perform pain scoring based on the physiological data set, facial expression data set, voice data set, the patient's basic information and self-assessment data, and generate a pain scoring result; S400: Receive the pain scoring result and the patient's self-controlled analgesia pump adjustment request, configure the safety threshold according to the patient's basic information, and use the pain scoring result, the self-controlled analgesia pump adjustment request and the safety threshold to generate a graded warning.

[0052] Among them, S300 includes:

[0053] S310: After receiving the facial expression data set, sort the facial expression data in chronological order, capture continuous changes in facial movements based on the sorting result, establish a first recognition result based on the result of capturing continuous changes in facial movements, obtain the patient's standard facial expression, compare the facial expression data set with the standard facial expression, establish a second recognition result, and feed the first recognition result and the second recognition result back to the storage center of the multidimensional pain assessment layer.

[0054] Among them, S300 includes:

[0055] S320: Perform heart rate variability and respiratory rate analysis on the physiological data set to generate a third recognition result, and feed the third recognition result back to the storage center of the multidimensional pain assessment layer.

[0056] Among them, S300 includes:

[0057] S330: After reading the voice data set, a fourth recognition result is established based on language keywords and intonation analysis, and the fourth recognition result is fed back to the storage center of the multidimensional pain assessment layer; S340: After reading the first recognition result, the second recognition result, the third recognition result, and the fourth recognition result stored in the storage center, a joint analysis is performed to establish a pain score result.

[0058] Among them, S310 includes:

[0059] S311: Establish a first recognition constraint and a second recognition constraint, wherein the first recognition constraint is a short-term burst recognition constraint, and the second recognition constraint is a continuously changing recognition constraint; S312: Use the first recognition constraint and the second recognition constraint to perform trigger analysis on the capture results of continuously changing facial movements, and perform weighted analysis of short-term bursts and continuously changing according to the trigger analysis results to establish the first recognition result.

[0060] Among them, S400 includes:

[0061] S410: extracting age information, weight information, and medical history information from the basic information; S420: performing threshold matching based on the age information, weight information, and medical history information, and performing threshold fallback and retention on the threshold matching result to establish the safety threshold.

[0062] Among them, S400 includes:

[0063] S430: Perform threshold trigger analysis on the automatic analgesia pump adjustment request according to the safety threshold, and generate a first warning according to the threshold trigger analysis result; S440: Perform pain warning identification according to the pain score result, establish a second warning, and generate a graded warning according to the first warning and the second warning.

[0064] Among them, S420 includes:

[0065] S421: establishing a pain coefficient according to the pain scoring result; S422: configuring a safety range base value based on the threshold matching result, and reconstructing a safety threshold through the pain coefficient and the safety range base value.

[0066] Among them, S400 includes:

[0067] When it is detected that the respiratory frequency monitored by the monitor of the multi-source device meets the preset frequency threshold, a coordinated stop instruction is generated, and the coordinated stop instruction is used to control the automatic analgesia pump to pause and issue an early warning.

[0068] An artificial intelligence-based full-process patient self-controlled analgesia management platform provided by an embodiment of the present invention can execute an artificial intelligence-based full-process patient self-controlled analgesia management method provided by any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

[0069] Although this application makes various references to certain modules in the platform according to the embodiments of this application, any number of different modules can be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other and are not used to limit the scope of protection of the present invention.

[0070] The above specific embodiments do not constitute a limitation to the scope of protection of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of this application should be included in the scope of protection of this application. In some cases, the actions or steps recorded in this application can be performed in an order different from that in the embodiments and can still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A patient-controlled analgesia management platform based on artificial intelligence throughout the entire process, characterized by: The platform includes: A multi-source data interaction module is used to access multi-source devices, read device data from multi-source devices, and establish a time series standard data set. The multi-source devices include monitors, patient-controlled analgesia pumps, and wearable devices; An information configuration module is used to configure the operating parameters of the automatic analgesia pump and then start the automatic analgesia pump by an operator with preset permissions; The AI pain assessment module is used to parse the updated time-series standard data set, extract the physiological data set, facial expression data set, and voice data set, and call the patient's basic information and self-assessment data. It activates the multidimensional pain assessment layer to perform pain scoring based on the physiological data set, facial expression data set, voice data set, the patient's basic information and self-assessment data, and generates the pain score result; An identification and warning module is used to receive the pain score result and the patient's patient-controlled analgesia pump adjustment request, configure a safety threshold based on the patient's basic information, and generate a graded warning using the pain score result, the patient-controlled analgesia pump adjustment request, and the safety threshold.

2. The artificial intelligence-based full-process patient-controlled analgesia management platform according to claim 1, characterized in that: The multidimensional pain assessment layer includes: The pain facial expression recognition channel is used to receive the facial expression data set, sort the facial expression data in chronological order, capture continuous changes in facial movements based on the sorting results, establish a first recognition result based on the continuous changes in facial movements captured results, obtain the patient's standard facial expression, compare the facial expression data set with the standard facial expression, establish a second recognition result, and feed back the first recognition result and the second recognition result to the storage center of the multidimensional pain assessment layer.

3. The artificial intelligence-based full-process patient-controlled analgesia management platform according to claim 2, characterized in that: In the pain facial expression recognition channel, establishing a first recognition result based on the result of continuously changing facial movements also includes: Establishing a first identification constraint and a second identification constraint, wherein the first identification constraint is a short-term burst identification constraint, and the second identification constraint is a continuously changing identification constraint; The first recognition constraint and the second recognition constraint are used to perform trigger analysis on the capture results of continuous changes in facial movements, and a weighted analysis of short-term bursts and continuous changes is performed based on the trigger analysis results to establish the first recognition result.

4. The artificial intelligence-based full-process patient-controlled analgesia management platform according to claim 3, characterized in that: The multidimensional pain assessment layer also includes: The physiological recognition channel is used to perform heart rate variability and respiratory rate analysis on the physiological data set, generate a third recognition result, and feed the third recognition result back to the storage center of the multidimensional pain assessment layer.

5. The artificial intelligence-based full-process patient-controlled analgesia management platform according to claim 4, characterized in that: The multidimensional pain assessment layer also includes: A language sentiment analysis channel is used to read the speech data set, analyze language keywords and intonation, establish a fourth recognition result, and feed the fourth recognition result back to the storage center of the multidimensional pain assessment layer; The comprehensive analysis channel is used to read the first recognition result, the second recognition result, the third recognition result, and the fourth recognition result stored in the storage center, perform joint analysis, and establish a pain score result.

6. The artificial intelligence-based full-process patient-controlled analgesia management platform according to claim 1, characterized in that: The identification and warning module includes: The information extraction submodule is used to extract age information, weight information, and medical history information from the basic information; The adaptive threshold setting module is used to perform threshold matching based on the age information, the weight information and the medical history information, and perform threshold backoff and retention on the threshold matching result to establish the safety threshold.

7. The artificial intelligence-based full-process patient-controlled analgesia management platform according to claim 6, characterized in that: The identification and warning module also includes: a comparison submodule, configured to perform a threshold trigger analysis on the patient-controlled analgesia pump adjustment request according to the safety threshold, and generate a first warning according to the threshold trigger analysis result; The pain warning submodule is used to perform pain warning identification based on the pain scoring result, establish a second warning, and generate a graded warning based on the first warning and the second warning.

8. The artificial intelligence-based full-process patient-controlled analgesia management platform according to claim 6, characterized in that: The adaptive threshold setting module is further used for: Establishing a pain index based on the pain score results; A safety range base value is configured based on the threshold matching result, and a safety threshold is reconstructed using the pain coefficient and the safety range base value.

9. The artificial intelligence-based full-process patient-controlled analgesia management platform according to claim 1, characterized in that: The platform also includes: The device collaboration module is used to generate a collaborative stop instruction when it detects that the monitored respiratory rate of the monitor of the multi-source device meets the preset frequency threshold, use the collaborative stop instruction to control the automatic control analgesia pump to pause, and issue an early warning.

10. A patient-controlled analgesia management method based on artificial intelligence throughout the entire process, characterized in that: The method is implemented by an artificial intelligence-based full-process patient-controlled analgesia management platform according to any one of claims 1 to 9, and the method comprises: Accessing multiple source devices, reading device data from the multiple source devices, and establishing a time series standard data set. The multiple source devices include monitors, patient-controlled analgesia pumps, and wearable devices; After configuring the operating parameters of the automatic analgesia pump, the operator with preset permissions will start the automatic analgesia pump; Parse the updated time series standard data set, extract the physiological data set, facial expression data set, and voice data set, call the patient's basic information and self-assessment data, activate the multidimensional pain assessment layer to perform pain scoring based on the physiological data set, facial expression data set, voice data set, the patient's basic information and self-assessment data, and generate the pain score result; The pain score result and the patient's patient-controlled analgesia pump adjustment request are received, and a safety threshold is configured according to the patient's basic information, and a graded warning is generated using the pain score result, the patient-controlled analgesia pump adjustment request and the safety threshold.

Images