Pain AI identification model training system and data storage device

Abstract

In the pain AI recognition model training system and the data storage device, an information collection assembly comprises a pain physiological information collection device and a subjective pain assessment device. The pain physiological information collection device is in signal communication with the AI training calculation assembly; the pain physiological information collection device collects pain-related physiological information of the testee, wherein the pain-related physiological information comprises heart rate information and electroencephalogram information; the subjective pain assessment device collects subjective pain assessment information of the testee; the information collection assembly sends the pain-related physiological information and the subjective pain assessment information to the AI training calculation assembly. After the AI training calculation assembly obtains the pain related physiological information and the subjective pain feeling evaluation information, AI training is carried out, and a pain feature data set is obtained after AI training. A pain AI recognition model training system is provided; objectively quantified physiological information and subjective pain assessment information can be collected, and a pain feature data set is established through an AI training calculation assembly.

Description

Technical Field

[0001] The present application belongs to the technical field of pain quantitative analysis, and specifically relates to a pain AI recognition model training system and a data storage device. Background Art

[0002] EEG information: This article specifically refers to the brain waves collected by EEG detection probes and the brain wave data formed after information extraction. The above brain wave data can be transmitted using various communication methods or stored and analyzed using computer technology.

[0003] Heart rate information: This article specifically refers to the heart beat information extracted through pulse detection, electrocardiogram detection or blood oxygen detection. The above heart beat information can be transmitted using various communication methods or stored and analyzed using computer technology.

[0004] Blood pressure information: This article specifically refers to the blood pressure information obtained through various technical means. The above blood pressure information can be transmitted using various communication methods or stored and analyzed using computer technology.

[0005] Pain is one of the most common clinical symptoms, and chronic pain is a disease. Currently, chronic pain is regarded as the third largest health problem after cardiovascular and cerebrovascular diseases and tumors. It not only causes physical discomfort to patients, but also leads to negative conditions such as insomnia, anxiety, and depression, which seriously affect the physical and mental health and quality of life of patients. Therefore, it is particularly important to formulate scientific and effective analgesic strategies, which can not only promote the recovery of patients' diseases and improve their quality of life, but also reduce economic and social burdens and optimize the allocation of medical resources. Effective treatment depends on accurate and objective assessment of pain status.

[0006] However, in the prior art, pain assessment can only rely on verbal reports from patients, such as visual analog scales, digital rating scales, etc. Such methods have many limitations in clinical application due to their strong subjectivity. This is especially difficult to apply to patients who cannot communicate (such as infants and patients with impaired consciousness). Therefore, there is an urgent need for a method that can objectively assess pain status.

[0007] Most of the current objective pain assessment methods rely on a single biological signal (such as EEG information), and fail to fully consider the potential impact of basic differences. At the same time, most of these methods focus on the assessment of laboratory-induced pain (such as electrical stimulation-induced pain), which lacks applicability in actual clinical applications.

[0008] Among the prior arts, the closest to the present invention is the pain detection and positioning method and system based on brain waves and neural networks. However, this system has many defects when using EEG information for pain assessment. For example, it is only applicable to evaluation within the laboratory, and does not consider the impact of basic differences in patients (age, gender, pain duration, pain site). The extraction range of EEG information signal features is small, and the large number of electrodes used leads to an increase in interference from irrelevant factors. Therefore, it cannot accurately and objectively reflect the patient's brain activity and pain perception process.

[0009] The use of AI (Artificial Intelligence) technology to analyze the relationship between EEG information and pain has been studied for many years. It was found that specific EEG information appears when pain occurs, but the accuracy is not high. The main reason is that the time of pain is relatively random. How to accurately collect EEG information when pain occurs is the key to whether AI can be successfully trained.

[0010] Although a large number of experiments have been able to collect EEG information at the moment of pain through external pain stimulation, this EEG information is associated with the pain at the external stimulation site. At the same time, the stimulation will also cause brain signal activity, interfering with the accuracy of AI training.

[0011] The technical problem solved by this application is how to collect pain EEG signals in daily life. Accurately collecting EEG signals when there is no interference or little interference stimulation to the tester is the basis for AI to identify various pains. Summary of the invention

[0012] The technical problem to be solved by this application is to avoid the shortcomings of the existing technology that cannot objectively evaluate pain, and propose a pain AI recognition model training system; it can collect physiological information data and subjective pain assessment information for objective quantification, and establish a pain feature data set through AI training calculation components. Let the correlation between objective physiological information data and subjective pain assessment information establish a deep relationship, and it can be continuously iterated. Based on this, the objectivity and accuracy of pain AI recognition will be improved.

[0013] The technical solution of the present application to solve the above-mentioned problems is a pain AI recognition model training system, including an information collection component and an AI training computing component; the information collection component includes a pain physiological information collection device and a subjective pain assessment device; the pain physiological information collection device is signal-connected to the AI ​​training computing component; the pain physiological information collection device collects pain-related physiological information of the tester, including heart rate information and EEG information; the number of the above-mentioned testers is greater than a set value; the subjective pain assessment device collects the subjective pain assessment information of the tester; the information collection component sends the above-mentioned pain-related physiological information and the subjective pain assessment information to the AI ​​training computing component; after the AI ​​training computing component obtains the above-mentioned pain-related physiological information and the above-mentioned subjective pain assessment information, it performs AI training, and obtains a pain feature data set after the AI ​​training.

[0014] The pain physiological information collection device includes a blood oxygen detection module, a heart rate detection module, and an electroencephalogram detection module; the above-mentioned heart rate detection module obtains the above-mentioned heart rate information based on the PPG signal obtained by the blood oxygen detection module, and the above-mentioned pain-related physiological information includes blood oxygen information; the electroencephalogram detection module is used to obtain the above-mentioned electroencephalogram information.

[0015] The above-mentioned pain AI recognition model training system includes any one of the following technical features: TC1: The above-mentioned subjective pain assessment device includes a pain start information input device. When the tester feels pain, he / she inputs the pain start information, and the pain physiological information collection device starts to collect pain-related physiological information; TC2: The above-mentioned subjective pain assessment device includes a pain start information input device. When the tester feels pain, he / she inputs the pain start information, and the pain physiological information collection device records the pain start time point; TC3: The above-mentioned subjective pain assessment device includes a pain end information input device. When the tester feels the pain is over, he / she inputs the pain end information, and the pain physiological information collection device records the pain end time point.

[0016] The subjective pain assessment device includes a pain intensity feedback device. When the test subject feels pain, the device feeds back pain intensity signals of different intensities according to the pain intensity.

[0017] The information collection component also includes a reference time device, which provides a time signal to the pain physiological information collection device and the subjective pain assessment device.

[0018] Information registration management terminal; the information collection component includes a network module, and the network module can be connected to the AI ​​training computing component; the information registration management terminal includes a Bluetooth module, and the above-mentioned pain physiological information collection device includes a Bluetooth module; the information registration management terminal is connected to the pain physiological information collection device via a Bluetooth signal; the above-mentioned subjective pain assessment device includes a Bluetooth module, and the subjective pain assessment device is connected to the pain physiological information collection device via a Bluetooth signal.

[0019] The above-mentioned pain AI recognition model training system includes any one of the following technical features: TA1: The above-mentioned AI training computing component collects pain-related physiological information, including blood oxygen signals, heart rate information, EEG information, and blood pressure information; TA2: The above-mentioned AI training computing component collects subjective pain assessment information, including pain start information, pain end information, and pain intensity information.

[0020] The above-mentioned pain AI recognition model training system includes any one of the following technical features: TB1: The above-mentioned AI training computing component collects tester information including the tester's age, gender, and pain attribute information; TB2: The above-mentioned AI training computing component collects tester information including the tester's age, gender, and pain attribute information; The pain attribute information includes any one of the pain site information and pain duration information.

[0021] The pain physiological information collection device uses a single-channel EEG acquisition system to collect the tester's scalp EEG information; the sampling rate is 1000 Hz; the length of each time period is greater than 2 seconds.

[0022] The technical solution of the present application to solve the above-mentioned problem can also be a data storage device for storing the pain feature data set involved in the above-mentioned pain AI recognition model training system.

[0023] The beneficial technical effects of the present application include: in the pain AI recognition model training system, a pain physiological information collection device and a subjective pain assessment device are set up to collect objective quantified physiological information data and subjective pain assessment information respectively, and a pain feature data set is established through an AI training calculation component.

[0024] The beneficial technical effects of this application include, with the help of the pain feature data set trained by AI, establishing a deep relationship between the correlation between objective physiological information data and subjective pain assessment information, and being able to iterate continuously. Based on this, the accuracy of pain AI recognition will be improved accordingly.

[0025] The beneficial technical effects of the present application include that the pain feature data set is obtained by the joint training of pain-related physiological information and subjective pain assessment information, and such an AI training process is more objective and realistic. Compared with the AI ​​training of pain features stimulated by external stimuli, the pain feature data set of the present application is closer to the real pain scene. It gets rid of the limitation that a stimulation device must be set up to train the pain AI recognition model. Data collection and training in more realistic scenarios are carried out in a freer scenario.

[0026] The beneficial technical effects of the present application include that the pain physiological information collection device includes a blood heart rate detection module and an electroencephalogram detection module; when pain occurs, the heart rate generally changes, and a correlation is established between the normal heart rate and the heart rate during pain. The AI ​​program can more easily detect the occurrence of pain and does not rely solely on brain wave information.

[0027] The beneficial technical effects of the present application include, that the pain physiological information collection device collects blood oxygen information; when pain occurs, whether the blood oxygen saturation changes, especially some chronic pain, the relationship between pain and blood oxygen, the relationship between psychological state and blood oxygen, and the relationship between blood oxygen and pain, which can be established through data collection and AI analysis to provide a basis for analysis.

[0028] The beneficial technical effects of the present application include: a blood oxygen detection device that can be easily made into a portable detection device, and can extract heart rate information when detecting the PPG signal in the blood oxygen; the system structure is simple and reliable, and as a portable device, it has low power consumption and can be worn for a long time.

[0029] The beneficial technical effects of the present application include: the pain physiological information collection device includes a blood oxygen detection module, a heart rate detection module, and an electroencephalogram detection module; the pain AI recognition model is trained based on multiple physiological parameters to enrich the dimensions of physiological data analysis and establish multi-dimensional comprehensive connections, so that the trained system is more objective and closer to reality when evaluating pain.

[0030] The beneficial technical effects of the present application include that the blood oxygen detection module and the heart rate detection module are easier to carry than other physiological parameters, and are more convenient for large-scale data collection and application.

[0031] The beneficial technical effects of the present application include the setting of the pain start information input device and the end button, synchronously starting the collection or identification of physiological information at the moment of pain start, and synchronously starting the collection or identification of physiological information at the moment of pain end, reducing the data load of the system, making it easier to grasp the key information related to pain, improving the efficiency of data collection and analysis, and avoiding the time, storage space and computing power consumption of searching for pain data in massive data for a long time.

[0032] The beneficial technical effects of the present application include that the pain intensity feedback device feeds back pain intensity signals of different intensities according to the pain intensity, and the pain intensity information provides a quantitative classification benchmark for evaluating the quantification of pain. The quantification and unification of pain intensity information can facilitate the long-term pain evaluation of individuals. The quantification and unification of pain intensity information can also facilitate comparative studies on pain evaluation of different individuals.

[0033] The beneficial technical effects of the present application include that the reference time device provides time signals to the pain physiological information collection device and the subjective pain assessment device, ensuring time synchronization between the two sets of devices, ensuring time consistency of signal processing, and improving the accuracy of the pain AI recognition model.

[0034] The beneficial technical effects of the present application include that the information registration management end provides a network module that can connect to the AI ​​training computing component, allowing the computing power distribution of the AI ​​training computing component to be more free, and the pain AI recognition model training can be carried out with the help of a larger external computing power platform.

[0035] The beneficial technical effects of the present application include that the information registration management terminal includes a Bluetooth module, and the pain physiological information collection device includes a Bluetooth module, which facilitates the collection of pain-related physiological information. Various signal connections at the data collection terminal are avoided, especially when multiple physiological information collection is carried out at the same time on site, and when data from different physiological information collection devices need to be collected, the wireless connection method greatly facilitates data transmission on site.

[0036] The beneficial technical effects of the present application include that the input of pain start information, pain end information, and pain intensity information provides basic information for converting subjective pain assessment into objective assessment, and establishes a channel for subjective and objective links for the quantification of pain.

[0037] The beneficial technical effects of the present application include that the AI ​​training computing component collects tester information including the tester's age, gender, and pain attribute information; the corresponding data can be classified and collected according to the tester's age, gender, and pain attribute information, allowing different groups of people to establish their own corresponding pain feature data sets, so that the pain feature data sets can more accurately reflect the pain attributes of different testers of different ages and genders.

[0038] The beneficial technical effects of the present application include that pain attribute information includes pain site information and pain duration, which further refines the pain differentiation mechanism, classifies and collects corresponding data for different pain sites and different pain durations, and establishes corresponding pain feature data sets for different pain sites and different pain durations, allowing the quantitative assessment of pain to go to a deeper level. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 Schematic diagram of pain AI recognition model training system Figure 1 ;

[0040] Figure 2 Schematic diagram of pain AI recognition model training system Figure 2 ;

[0041] Figure 3 Schematic diagram of the subjective pain assessment device Figure 1 ;

[0042] Figure 4 Schematic diagram of the subjective pain assessment device Figure 2 ;

[0043] Figure 5 Schematic diagram of pain AI recognition model training system Figure 3 ;

[0044] Figure 6 Schematic diagram of pain AI recognition model training system Figure 4 ;

[0045] Figure 7 Schematic diagram of pain AI recognition model training system Figure 5 ;

[0046] Figure 8 Schematic diagram of pain AI recognition model training system Figure 6 ;

[0047] Fig. 9 Schematic diagram of pain AI recognition model training system Figure 7 ;

[0048] Fig.10 Schematic diagram of pain information collection of subjective pain assessment device;

[0049] Fig.11 Schematic diagram of pain AI recognition model training system Figure 8 . DETAILED DESCRIPTION

[0050] The implementation methods of the present application are further described in detail below in conjunction with the accompanying drawings.

[0051] In this application, in the pain AI recognition model training system based on EEG information, through the information collection component and the AI ​​training calculation component and taking into account the basic information of the patient, the pain AI recognition model training can be used to obtain a pain feature data set for AI recognition of pain. Based on the pain AI recognition model training system, a more objective and accurate pain assessment model can be obtained. This system will enhance the applicability and practicality of pain AI recognition in clinical environments and achieve more scientific and effective pain management. By fusing multidimensional information, an assessment model that can evaluate the patient's pain status in real time is established, thereby achieving real-time and accurate assessment of the pain status.

[0052] like Figure 1 and Figure 2In an embodiment of a pain AI recognition model training system, the system includes an information collection component and an AI training computing component; the information collection component includes a pain physiological information collection device and a subjective pain assessment device; the pain physiological information collection device is signal-connected to the AI ​​training computing component; the pain physiological information collection device collects pain-related physiological information of the tester, and the pain-related physiological information includes blood oxygen signal, heart rate information, and electroencephalogram information; the subjective pain assessment device collects the subjective pain assessment information of the tester; the information collection component sends the above-mentioned pain-related physiological information and subjective pain assessment information to the AI ​​training computing component; the number of the above-mentioned testers is greater than a set value; according to the pain-related physiological information and subjective pain assessment information of the set number of testers, the AI ​​training computing component calculates and obtains a pain feature data set.

[0053] like Figure 2 In an embodiment of a pain AI recognition model training system, the pain physiological information collection device includes a blood oxygen detection module, a heart rate detection module, and an electroencephalogram detection module; the heart rate detection module obtains heart rate information based on the PPG signal obtained by the blood oxygen detection module.

[0054] like Figure 3 In an embodiment of a pain AI recognition model training system, the above-mentioned subjective pain assessment device includes a pain start information input device. When the tester feels pain, the pain start information is input, and the pain physiological information collection device starts to collect pain-related physiological information; the above-mentioned subjective pain assessment device includes a pain end information input device. When the tester feels the pain is over, the pain end information is input, and the pain physiological information collection device records the pain end time point.

[0055] like Figure 3 In an embodiment of a pain AI recognition model training system, the above-mentioned subjective pain assessment device includes a pain start information input device. When the tester feels pain, he / she inputs the pain start information, and the pain physiological information collection device records the pain start time point; the above-mentioned subjective pain assessment device includes a pain end information input device. When the tester feels the pain ends, he / she inputs the pain end information, and the pain physiological information collection device marks the pain end mark on the collected pain-related physiological information.

[0056] like Figure 3 In an embodiment of a pain AI recognition model training system, the subjective pain assessment device includes a pain intensity feedback device. When the tester feels pain, pain intensity signals of different intensities are fed back according to the pain intensity.

[0057] like Figure 6 In an embodiment of a pain AI recognition model training system, the information collection component also includes a reference time device, which provides a time signal to the pain physiological information collection device and the subjective pain assessment device.

[0058] like Figure 4 In an embodiment of a pain AI recognition model training system, the above-mentioned subjective pain assessment device includes a Bluetooth module, and the subjective pain assessment device is connected to the pain physiological information collection device via a Bluetooth signal.

[0059] like Figure 7 and Fig. 9 In an embodiment of a pain AI recognition model training system, the information collection component includes a network module, and the information collection component can connect to the AI ​​training computing component and the information registration management terminal through the network module.

[0060] like Figure 8 In an embodiment of a pain AI recognition model training system, the information registration management end includes a Bluetooth module, and the above-mentioned information collection component includes a Bluetooth module; data is transmitted between the information registration management end and the information collection component via a Bluetooth connection.

[0061] like Fig.11 In an embodiment of the pain AI recognition model training system, the pain physiological information collection device includes a Bluetooth module; the information registration management terminal is connected to the pain physiological information collection device via a Bluetooth signal; the subjective pain assessment device includes a Bluetooth module, and the subjective pain assessment device is connected to the pain physiological information collection device via a Bluetooth signal. The subjective pain assessment device can be a portable client, such as a mobile phone. The pain physiological information collection device can also be a portable signal ointment device, such as a bracelet. The two can also be integrated into an integrated device.

[0062] like Figure 4 and Figure 5 In an embodiment of a pain AI recognition model training system, the AI ​​training computing component collects pain-related physiological information including blood oxygen signals, heart rate information, EEG information, and blood pressure information; the AI ​​training computing component collects subjective pain assessment information including pain start information, pain end information, and pain intensity information. Fig.10 ,Subjective pain assessment device, is used to collect pain onset information, pain end information, and pain intensity information.

[0063] The above-mentioned AI training computing component collects tester information including the tester's age, gender, and pain attribute information; the above-mentioned AI training computing component collects tester information including the tester's age, gender, and pain attribute information; the pain attribute information includes any one of pain site information and pain duration information.

[0064] The pain physiological information collection device uses a single-channel EEG acquisition system to collect the tester's scalp EEG information; the sampling rate is 1000 Hz; the length of each time period is greater than 2 seconds.

[0065] A data storage device for storing the pain feature data set involved in the above-mentioned pain AI recognition model training system.

[0066] In the pain AI recognition model training method based on the pain AI recognition model training system, the specific steps are as follows:

[0067]

S1

[0068]

S2

[0069]

S3

[0070]

S4

[0071]

S5

[0072]

S6

[0073]

S7

[0074]

S8

[0075]

S9

[0076]

S10

[0077]

S11

[0078]

S12

[0079]

S13

[0080] By collecting basic information about patients and using a single-channel EEG system to collect brain activity data, signal processing technology and machine learning algorithms are combined to perform feature extraction and pattern recognition on the EEG information, ultimately generating pain assessment results.

[0081] Objectivity: By directly monitoring brain activity, the influence of subjective factors is reduced. Real-time: Real-time collection and analysis of EEG information to quickly provide pain assessment results. Personalization: Combined with the patient's individual information, personalized pain assessment is achieved. Pain assessment method based on EEG information, including technical implementation of signal acquisition, processing and analysis. Signal acquisition device using single-channel electrodes and its wireless transmission module. Technical solution for analyzing patient basic information and EEG information and pain assessment in combination with machine learning algorithms. Design and implementation of real-time display and alarm modules.

[0082] The serial numbers such as first and second in this application are only for the convenience of expression and do not necessarily indicate the order relationship in terms of size and time sequence. The letter serial numbers of the steps are also only for the convenience of expression and do not necessarily indicate the order relationship in terms of time sequence.

[0083] As shown in the accompanying drawings, the above are only embodiments of the present application, and the patent scope of the present application is not limited thereto. Any equivalent structure or equivalent process transformation made using the contents of the invention description and the accompanying drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A pain AI recognition model training system, characterized by: Including information collection components and AI training and computing components; The information collection component includes a pain physiological information collection device and a subjective pain assessment device; The pain physiological information collection device is connected to the AI ​​training computing component by signal; The pain physiological information collection device collects the tester's pain-related physiological information, including heart rate information and EEG information; The number of testers is greater than a set value; The subjective pain assessment device collects the tester's subjective pain assessment information; The information collection component sends the pain-related physiological information and the subjective pain assessment information to the AI ​​training calculation component; After the AI ​​training computing component obtains the pain-related physiological information and the subjective pain assessment information, AI training is performed to obtain a pain feature data set.

2. The pain AI recognition model training system according to claim 1, characterized in that: The pain physiological information collection device includes a blood oxygen detection module, a heart rate detection module, and an electroencephalogram detection module; the heart rate detection module obtains the heart rate information based on the PPG signal obtained by the blood oxygen detection module, and the pain-related physiological information includes blood oxygen information; the electroencephalogram detection module is used to obtain the electroencephalogram information.

3. The pain AI recognition model training system according to claim 1, characterized in that: Include any of the following technical features: TC1: The subjective pain assessment device includes a pain onset information input device. When the tester feels pain, the pain onset information is input, and the pain physiological information collection device starts to collect pain-related physiological information; TC2: The subjective pain assessment device includes a pain onset information input device. When the tester feels pain, he inputs the pain onset information, and the pain physiological information collection device records the pain onset time point; TC3: The subjective pain assessment device includes a pain end information input device. When the test subject feels that the pain has ended, he / she inputs the pain end information, and the pain physiological information collection device records the time point when the pain has ended.

4. The pain AI recognition model training system according to claim 1, characterized in that: The subjective pain assessment device includes a pain intensity feedback device. When the test subject feels pain, the device feeds back pain intensity signals of different intensities according to the pain intensity.

5. The pain AI recognition model training system according to claim 1, characterized in that: The information collection component also includes a reference time device, which provides a time signal to the pain physiological information collection device and the subjective pain assessment device.

6. The pain AI recognition model training system according to claim 1, further comprising: Information registration management terminal; the information collection component includes a network module, and the network module can be connected to the AI ​​training computing component; the information registration management terminal includes a Bluetooth module, and the pain physiological information collection device includes a Bluetooth module; The information registration management terminal is connected to the pain physiological information collection device via a Bluetooth signal; The subjective pain assessment device comprises a Bluetooth module, and the subjective pain assessment device is connected to the pain physiological information collection device via a Bluetooth signal.

7. The pain AI recognition model training system according to claim 6, characterized in that: Including any of the following technical features: TA1: The AI ​​training computing component collects pain-related physiological information, including blood oxygen signals, heart rate information, EEG information, and blood pressure information; TA2: The AI ​​training computing component collects subjective pain assessment information, including pain start information, pain end information, and pain intensity information.

8. The pain AI recognition model training system according to claim 6, characterized in that: Including any of the following technical features: TB1: The AI ​​training computing component collects tester information including the tester's age, gender, and pain attribute information; TB2: The AI ​​training computing component collects tester information including the tester's age, gender, and pain attribute information; The pain attribute information includes any one of pain site information and pain duration information.

9. The pain AI recognition model training system according to claim 1, characterized in that: The pain physiological information collection device uses a single-channel EEG acquisition system to collect the tester's scalp EEG information; the sampling rate is 1000 Hz; the length of each time period is greater than 2 seconds.

10. A data storage device, characterized in that: Used to store the pain feature data set involved in the pain AI recognition model training system described in any one of claims 1 to 8.

Images