A wearable device and recommendation system based on EEG and fNIRS

Abstract

The present invention discloses a wearable device and recommendation system based on EEG and fNIRS, relating to the field of bio-intelligent sensing technology. A signal acquisition module collects a user's specific physiological parameters; the specific physiological parameters include brain wave data and near-infrared light data; a data processing module preprocesses the specific physiological parameters to obtain preprocessed specific physiological parameters; a feature analysis module extracts individual scenario features from the preprocessed specific physiological parameters to obtain behavioral activity features and emotional state features; a rhythm synthesis module is configured to: identify and match stored musical rhythms using a feature network to obtain a first musical rhythm that matches the emotional state features; identify and match the first musical rhythm using a decision tree to obtain a second musical rhythm that matches the behavioral activity features; and a personalized recommendation module plays the second musical rhythm to the user. This invention effectively integrates EEG, fNIRS, and music recommendations.

Description

Technical Field

[0001] The present invention relates to the field of biological intelligent sensing technology, and in particular to a wearable device and a recommendation system based on EEG and fNIRS. Background Art

[0002] Electroencephalogram (EEG) detects bioelectric signals, which are derived from the postsynaptic potentials generated synchronously by a large number of neurons. It generally refers to the electrical signals that spread from the electrophysiological activity information of the cerebral cortex through the skull and diffuse to the scalp.

[0003] Functional near-infrared spectroscopy (fNIRS) detects optical signals derived from changes in optical properties caused by the rise and fall of oxyhemoglobin (HbO) and deoxyhemoglobin (Hb) concentrations in active brain tissue. These signals generally refer to the intensity of light scattered by the cerebral cortex, as blood has very low absorption of near-infrared light (700-900nm) and has good scattering properties.

[0004] Brain-computer interface technology based on EEG and fNIRS can realize the information transmission of controlling peripheral devices using brain information and can be applied in multiple fields.

[0005] Currently, music recommendations are mostly based on the user's previous music, recommending music of the same or similar genres. EEG, fNIRS, and music recommendations cannot be effectively combined. Summary of the Invention

[0006] The purpose of the embodiments of the present invention is to provide a wearable device and recommendation system based on EEG and fNIRS, which effectively combines EEG, fNIRS, and music recommendation.

[0007] To achieve the above objectives, the present invention provides the following solutions:

[0008] A wearable device based on EEG and fNIRS, comprising:

[0009] A signal acquisition module is used to collect specific physiological parameters of the user; the specific physiological parameters include: brain wave data and near-infrared light data;

[0010] A data processing module, configured to pre-process the specific physiological parameters and generate pre-processed specific physiological parameters;

[0011] A feature analysis module is used to extract individual scenario features from the pre-processed specific physiological parameters to obtain behavioral activity features and emotional state features;

[0012] Temperament synthesis module, used for:

[0013] Identify and match the stored music rhythms through a feature network to obtain a first music rhythm that matches the emotional state feature;

[0014] Identify and match the first music rhythm through a decision tree to obtain a second music rhythm that matches the behavioral activity characteristics;

[0015] A personalized recommendation module is used to play the second music rhythm to the user.

[0016] Optionally, the process of constructing the feature network is specifically as follows:

[0017] Collecting and analyzing the behavioral activity characteristics and emotional state characteristics of multiple users; forming the behavioral activity characteristics of the multiple users into a behavioral activity characteristic set; wherein one behavioral activity characteristic set includes multiple different behavioral activity characteristic subsets;

[0018] Combining the emotional state characteristics of multiple users into an emotional state characteristic set; one of the emotional state characteristic sets includes multiple different emotional state characteristic subsets;

[0019] Classifying the stored music rhythms to obtain a plurality of first music rhythms; matching the plurality of first music rhythms with the plurality of different emotional state feature subsets;

[0020] The first music rhythm is classified to obtain a plurality of second music rhythms; the plurality of second music rhythms are matched with the plurality of different behavioral activity feature subsets.

[0021] Optionally, identifying and matching the first music rhythm by using a decision tree specifically includes:

[0022] The decision tree includes a plurality of sub-decision trees; the plurality of sub-decision trees correspond one-to-one to the plurality of different emotional state feature subsets; the plurality of different emotional state feature subsets match the plurality of second music rhythms;

[0023] Inputting the plurality of different emotional state feature subsets into the decision tree to obtain corresponding sub-decision trees;

[0024] The corresponding sub-decision tree decides a matching second music rhythm.

[0025] Optionally, the signal acquisition module includes:

[0026] An EEG acquisition unit, configured to acquire the brain wave data;

[0027] fNIRS acquisition unit, used for acquiring the near-infrared light data;

[0028] A synchronous acquisition unit is connected to the EEG acquisition unit and the fNIRS acquisition unit respectively, and is used to synchronously acquire the brain wave data and the near-infrared light data to obtain synchronized brain wave data and near-infrared light data.

[0029] Optionally, the data processing module includes:

[0030] a filtering unit connected to the synchronization acquisition unit, and configured to filter the synchronized brain wave data and near-infrared light data to obtain filtered brain wave data and near-infrared light data;

[0031] an artifact recognition unit connected to the filtering unit, configured to: identify and delete data having a voltage value greater than a first threshold value in the filtered EEG data, and identify and delete data having a fluctuation deviation value greater than a second threshold value in the near-infrared light data, thereby obtaining artifact-recognized EEG data and near-infrared light data;

[0032] The baseline calibration unit is connected to the artifact recognition unit and is used to perform a third threshold calibration on the brain wave data and near-infrared light data after artifact recognition in terms of duration to obtain calibrated brain wave data and near-infrared light data.

[0033] To achieve the above objectives, the present invention further provides the following solutions:

[0034] A recommendation system based on EEG and fNIRS, including:

[0035] A first signal acquisition module is used to collect specific physiological parameters of the user; the specific physiological parameters include: brain wave data and near-infrared light data;

[0036] Cloud servers for:

[0037] Preprocessing the specific physiological parameters to form preprocessed specific physiological parameters;

[0038] Extracting individual situational features from the pre-processed specific physiological parameters to obtain behavioral activity features and emotional state features;

[0039] Identify and match the stored music rhythms through a feature network to obtain a first music rhythm that matches the emotional state feature;

[0040] Identify and match the first music rhythm through a decision tree to obtain a second music rhythm that matches the behavioral activity characteristics;

[0041] The first personalized recommendation module is used to play the second music rhythm to the user.

[0042] Optionally, the process of constructing the feature network is specifically as follows:

[0043] Collect and analyze the behavioral activity characteristics and emotional state characteristics of multiple users;

[0044] Combining the behavioral activity features of multiple users into a behavioral activity feature set; one behavioral activity feature set includes multiple different behavioral activity feature subsets;

[0045] Combining the emotional state characteristics of multiple users into an emotional state characteristic set; one of the emotional state characteristic sets includes multiple different emotional state characteristic subsets;

[0046] Classifying the stored music rhythms to obtain a plurality of first music rhythms; matching the plurality of first music rhythms with the plurality of different emotional state feature subsets;

[0047] The first music rhythm is classified to obtain a plurality of second music rhythms; the plurality of second music rhythms are matched with the plurality of different behavioral activity feature subsets.

[0048] Optionally, identifying and matching the first music rhythm by using a decision tree specifically includes:

[0049] The decision tree includes a plurality of sub-decision trees; the plurality of sub-decision trees correspond one-to-one to the plurality of different emotional state feature subsets; the plurality of different emotional state feature subsets match the plurality of second music rhythms;

[0050] Inputting the plurality of different emotional state feature subsets into the decision tree to obtain corresponding sub-decision trees;

[0051] The corresponding sub-decision tree decides a matching second music rhythm.

[0052] Optionally, the first signal acquisition module includes:

[0053] An EEG acquisition unit, configured to acquire the brain wave data;

[0054] fNIRS acquisition unit, used for acquiring the near-infrared light data;

[0055] A synchronous acquisition unit is connected to the EEG acquisition unit and the fNIRS acquisition unit respectively, and is used to synchronously acquire the brain wave data and the near-infrared light data to obtain synchronized brain wave data and near-infrared light data.

[0056] Optionally, the cloud server includes:

[0057] a filtering unit connected to the synchronization acquisition unit, and configured to filter the synchronized brain wave data and near-infrared light data to obtain filtered brain wave data and near-infrared light data;

[0058] an artifact recognition unit connected to the filtering unit, configured to: identify and delete data having a voltage value greater than a first threshold value in the filtered EEG data, and identify and delete data having a fluctuation deviation value greater than a second threshold value in the near-infrared light data, thereby obtaining artifact-recognized EEG data and near-infrared light data;

[0059] The baseline calibration unit is connected to the artifact recognition unit and is used to perform a third threshold calibration on the brain wave data and near-infrared light data after artifact recognition in terms of duration to obtain calibrated brain wave data and near-infrared light data.

[0060] It can be seen that in the embodiment of the present invention, the signal acquisition module collects the user's specific physiological parameters; the specific physiological parameters include brain wave data and near-infrared light data. The advantages of brain wave data and near-infrared light data complement each other, so that the signal acquisition module can collect stronger signals, more accurate positioning, and stronger anti-interference ability.

[0061] The data processing module preprocesses the specific physiological parameters and forms the preprocessed specific physiological parameters for subsequent calculation.

[0062] The feature analysis module extracts individual contextual features from preprocessed specific physiological parameters, generating behavioral and emotional state features. The rhythm synthesis module uses a feature network to identify and match stored musical rhythms, generating a first rhythm that matches the emotional state features. A decision tree is then used to identify and match the first rhythm, generating a second rhythm that matches the behavioral features. The personalized recommendation module plays the second rhythm to the user. This enables bidirectional transmission of brain-computer signals, enabling wearable devices based on EEG and fNIRS to cover both health monitoring and management, integrating both detection and intervention. Combining the advantages of EEG, fNIRS, and music recommendation, this system expands the practical applications of wearable devices and reduces their production costs.

[0063] A recommendation system based on EEG and fNIRS uses cloud servers to access the Internet of Things, expanding the span of usage scenarios and time, greatly simplifying the collection and processing of user-specific physiological parameters, and increasing database storage capacity in scientific research and clinical fields.

[0064] A wearable device and recommendation system based on EEG and fNIRS performs real-time analysis of a large number of user-specific physiological parameters, enabling intelligent processing of music synthesis and recommended playback. This can provide targeted pain relief for users and better analyze and understand the emotional soothing effects of music.

[0065] Therefore, the present invention realizes the effective combination of EEG, fNIRS and music recommendation. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0067] Figure 1 A schematic diagram of the structure of a wearable device based on EEG and fNIRS provided in an embodiment of the present invention;

[0068] Figure 2 A schematic diagram of the structure of the EEG and fNIRS-based recommendation system provided in an embodiment of the present invention;

[0069] Figure 3 A schematic diagram of the decision tree model structure provided by an embodiment of the present invention;

[0070] Figure 4 This is a schematic structural diagram of a ring-shaped headband provided by an embodiment of the present invention.

[0071] Explanation of symbols:

[0072] Signal acquisition module-1, data processing module-2, feature analysis module-3, music synthesis module-4, personalized recommendation module-5, first signal acquisition module-6, cloud server-7, first personalized recommendation module-8. DETAILED DESCRIPTION

[0073] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0074] Providing timely music recommendations tailored to different users or groups of users, tailored to specific scenarios, has broad implications. For users experiencing high levels of psychological stress or engaging in intense mental work, music recommendations can help relieve tension, alleviate fatigue, and restore physical function. For those with mood disorders, numerous case studies have demonstrated that music recommendations can effectively alleviate negative emotions such as mania, bipolar disorder, and depression. For the elderly, music recommendations can enhance memory and cognitive function, reducing the risk of neurasthenia. For younger individuals, especially those with limited language skills, music recommendations are effective in improving verbal fluency. For those experiencing chronic pain, music recommendations can effectively reduce pain-related anxiety and improve sleep quality. Furthermore, music recommendations are easy to implement and, as a non-invasive method, can effectively alleviate emotional distress. Therefore, expanding the social benefits of music recommendations is beneficial to human health and development.

[0075] Although the role of music recommendation in relieving pain, releasing stress, and maintaining mental health has been gradually developed, its function is usually manifested in a non-standard environment without accurate objective basis. It relies solely on the user's subjective feelings to judge the individual's current physiological and psychological condition, and the corresponding music rhythm that matches it lacks factual basis and cannot promote the subsequent development of this field. Therefore, it is necessary to collect the user's specific physiological parameters in real time and conveniently, use specific physiological parameters as data support, present objective physiological indicators, and generate or recommend appropriate music rhythms according to personalized physiological scenarios, which will help music recommendations move towards more standardized research and personalized promotion in the market direction.

[0076] In addition, although a large number of behavioral studies have confirmed that music recommendations have obvious intervention effects, due to the variability of music itself and the diversity of user groups, the mechanism of how music recommendations affect users is still unclear. Therefore, combined with brain cognitive imaging technology, the effects of music therapy on the brain structure changes caused by various user groups are studied in big data. In the process of effectively utilizing music therapy, the mechanism framework behind music recommendation is continuously explored to lay a research foundation for the sustainable development of music therapy.

[0077] While EEG and fNIRS-based brain cognitive imaging technologies continue to advance, and smart wearable devices are also gaining momentum, there's a lack of effective integration between the two. While these technologies have yielded impressive experimental results, their reach hasn't been effectively extended to individual users. As smart wearables enter the consumer market, the amount of data they retain holds enormous potential, but it's struggled to be effectively utilized and interpreted. Therefore, combining smart wearables, EEG, and fNIRS allows for this data to be scientifically tested and analyzed, parallelized into automated data models. This allows scientific research to be applied to the consumer market, while also allowing retained data from the broader market to feed back into scientific research.

[0078] The purpose of the present invention is to provide a wearable device and recommendation system based on EEG and fNIRS to solve the problem that the existing EEG, fNIRS and music recommendation cannot be effectively combined.

[0079] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0080] Figure 1 The exemplary structure of the aforementioned wearable device based on EEG and fNIRS is shown, which includes at least a signal acquisition module 1, a data processing module 2, a feature analysis module 3, a music synthesis module 4, and a personalized recommendation module 5. Each module is described in detail below.

[0081] The signal acquisition module 1 is used to collect the user's specific physiological parameters; the specific physiological parameters include: brain wave data and near-infrared light data.

[0082] In one example, the specific physiological parameters are bioelectrical signals and near-infrared light signals emitted by the user's brain in real time. Signal acquisition module 1 collects the bioelectrical signals to produce brainwave data. Signal acquisition module 1 also collects the near-infrared light signals to produce near-infrared light data. Signal acquisition module 1 can collect specific physiological parameters in a laboratory or professional medical setting, or even at home without professional supervision, adapting to a variety of collection environments.

[0083] The data processing module 2 is used to preprocess the specific physiological parameters and form preprocessed specific physiological parameters.

[0084] In one example, after the specific physiological parameters enter the data processing module 2, the data processing module 2 mainly performs conventional data filtering, artifact identification, baseline calibration and other processing on the specific physiological parameters, and then outputs the pre-processed specific physiological parameters. The data processing module 2 not only supports the export of data interfaces, but also can transmit data in real time via Bluetooth, Wi-Fi, etc., and directly connects the pre-processed specific physiological parameters to the user's mobile device such as a mobile phone, making it convenient for the user to choose the presentation form of the pre-processed specific physiological parameters on their mobile device, such as brain area activation images, hemodynamic curves, real-time raw data, etc. The specific physiological parameters also include the user's demographic information, including age, gender, location, etc.

[0085] The feature analysis module 3 is used to extract individual situational features from the pre-processed specific physiological parameters to obtain behavioral activity features and emotional state features. Individual situational features include behavioral activity features and emotional state features.

[0086] In one example, the feature analysis module 3 can specifically be a convolutional neural network (CNN), which can also be called a feature network. After the pre-processed specific physiological parameters are input into the feature network, the feature network will extract individual scenario features based on the frequency parameters in the pre-processed specific physiological parameters. At this time, the feature network is a feature network that follows the individual user. Individual scenario features include the user's behavioral activity features and emotional state features. Behavioral activity features can be the frequency of behavioral rhythms such as running, walking, sitting still, and sleeping. Emotional state features include frequency band information contained in emotional states such as insomnia, depression, headache, and anxiety. When a user group is formed by multiple individual users, the feature network at this time is a feature network that follows the user group.

[0087] Behavioral activity characteristics not only include frequency but also include oxygen consumption, blood oxygen saturation, heart rate, temperature, and blood flow velocity during states like running, walking, sitting, and sleeping. Brain waves represent the electrophysiological activity of the cerebral cortex and can be used to infer the user's emotional state, thereby generating emotional state characteristics. Behavioral activity characteristics include metabolic indicators and the intensity of behavioral movements in different states. Emotional state characteristics also include the onset time and intensity of emotions such as insomnia, depression, headaches, and anxiety.

[0088] The rhythm synthesis module 4 is used for:

[0089] The stored music rhythms are identified and matched through the feature network to obtain the first music rhythm that matches the emotional state characteristics.

[0090] The first music rhythm is identified and matched through a decision tree to obtain a second music rhythm that matches the behavioral activity characteristics.

[0091] In one example, stored music rhythms can be pre-stored in a wearable device or updated in real time via the internet. Different music rhythms correspond to frequencies contained in different behavioral activity characteristics and emotional state characteristics. The emotional state characteristics and the stored music rhythms are simultaneously input into a feature network. The feature network matches a corresponding first music rhythm based on the frequencies contained in the emotional state characteristics, i.e., a first music rhythm that matches the emotional state characteristics. During the matching process, a small portion of the stored music rhythms can be matched first, and then all of the stored music rhythms can be matched. Each time the behavioral activity characteristics and the first music rhythm are simultaneously input into a decision tree, the decision tree matches a corresponding second music rhythm based on the frequencies contained in the behavioral activity characteristics, i.e., a second music rhythm that matches the behavioral activity characteristics. Alternatively, the behavioral activity characteristics and the first music rhythm can be simultaneously input into different decision trees. Different decision trees match different or the same corresponding second music rhythms based on the frequencies contained in the behavioral activity characteristics. The decision forest includes different decision trees. Behavioral activity characteristics include at least movement frequency, breathing rate, heart rate, vocalization frequency, etc.

[0092] The stored music rhythms are screened and classified according to acoustic and medical principles. The music rhythm that is suitable for the user is processed by medical acoustic matching. The matching process specifically includes: first, using PCA (principal component analysis) to reduce the dimensions of a large number of stored music rhythm parameters into a small number of parameters, including timbre, pitch, and frequency. The parameters are then centralized to obtain a covariance matrix. The covariance matrix is ​​then decomposed to obtain eigenvectors. The eigenvectors are arranged from large to small, and the first 5 eigenvectors are taken. The music rhythms corresponding to the 5 eigenvectors are the music rhythms that need to be stored. For example, the frequency range of general musical instruments is 20Hz~20kHz, while the highest sound heard by humans is about 15kHz.

[0093] There's a mapping relationship between the frequencies contained in the behavioral activity features and the second musical rhythm, which can be used to fit a decision tree. When the frequencies contained in the behavioral activity features are complex and exhibit a non-monotonic trend, a decision forest can be used.

[0094] In the embodiment of the present invention, the decision tree algorithm is suitable for data with high flexibility such as music recommendation. According to the following decision tree model, see Figure 3. First, combine the group attributes of the known user group and the pre-entered individual needs. The demand information is selected by the user to obtain the user's condition, that is, the emotional state characteristics mentioned above. Conditions include: insomnia, depression, headache, anxiety, etc. Then match the first music rhythm and the scene the user is currently in, including family, work, entertainment, etc. Get the individual state of the user, and then integrate the user's individual habits and individual goals to get the second music rhythm. Individual goals include: relieving headaches, relieving depression, relieving tension, etc. Individual goals are pre-set and selected by the user. The above forms an overall description of the individual user, which serves as the root of the decision tree. Different decision trees are equivalent to different overall descriptions of individual users, so the roots can be different.

[0095] The personalized recommendation module 5 is used to play the second music rhythm to the user.

[0096] In one example, the personalized recommendation module 5 may be a player or a speaker, as long as it can play the second music rhythm.

[0097] In the embodiments of the present invention, see Figure 4 The above-mentioned signal acquisition module 1, data processing module 2, feature analysis module 3, music synthesis module 4 and personalized recommendation module 5 can be designed as a ring-shaped headband in appearance, or can be designed into other shapes, as long as it can be flexibly adjusted for the user to wear.

[0098] In summary, the signal acquisition module 1, data processing module 2, feature analysis module 3, music synthesis module 4, and personalized recommendation module 5 enable bidirectional transmission of brain-computer signals, enabling wearable devices based on EEG and fNIRS to cover human health monitoring and management, integrating both detection and intervention. Combining the advantages of EEG, fNIRS, and music recommendation, this system enriches the practical applications of wearable devices and reduces their production costs.

[0099] The specific process of building a feature network is as follows:

[0100] Collect and analyze the behavioral activity characteristics and emotional state characteristics of multiple users.

[0101] For the specific acquisition process, please refer to the description of the signal acquisition module 1 above, which will not be repeated here.

[0102] The behavioral activity features of multiple users are combined into a behavioral activity feature set; a behavioral activity feature set includes multiple different behavioral activity feature subsets.

[0103] In one example, the behavioral activity features of multiple users constitute a behavioral activity feature set, and the multiple different behavioral activity feature subsets include: a running behavioral activity feature subset, a walking behavioral activity feature subset, a sitting behavioral activity feature subset, a sleeping behavioral activity feature subset, and the like.

[0104] The emotional state features of multiple users are combined into an emotional state feature set; an emotional state feature set includes multiple different emotional state feature subsets.

[0105] In one example, the emotional state features of multiple users constitute an emotional state feature set, and the multiple different emotional state feature subsets include: an insomnia emotional state feature subset, a depression emotional state feature subset, a headache emotional state feature subset, an anxiety emotional state feature subset, and the like.

[0106] The stored music rhythms are classified to obtain a plurality of first music rhythms; the plurality of first music rhythms are matched with a plurality of different emotional state feature subsets.

[0107] In one example, a mapping relationship exists between multiple first music rhythms and multiple different emotional state feature subsets. For example, the label value (intensity level) of the insomnia emotional state feature subset in the feature network is 30, the label value of the depression emotional state feature subset in the feature network is 40, the label value of the headache emotional state feature subset in the feature network is 50, the label value of the anxiety emotional state feature subset in the feature network is 60, and so on. Users are divided into corresponding subsets based on their selected individual goals.

[0108] The first music rhythm is classified to obtain a plurality of second music rhythms; the plurality of second music rhythms are matched with a plurality of different behavioral activity feature subsets.

[0109] In one example, there is a mapping relationship between the plurality of second music rhythms and the plurality of different behavioral activity feature subsets.

[0110] Identifying and matching the first music rhythm through the decision tree specifically includes:

[0111] The decision tree includes multiple sub-decision trees; the multiple sub-decision trees correspond one-to-one to multiple different emotional state feature subsets; and the multiple different emotional state feature subsets match multiple second music rhythms.

[0112] Multiple subsets of different emotional state features are input into the decision tree to obtain corresponding sub-decision trees.

[0113] The corresponding sub-decision tree decides the matching second music rhythm.

[0114] In one example, there is a mapping relationship between multiple sub-decision trees and multiple different emotional state feature subsets, and there is a mapping relationship between multiple different emotional state feature subsets and multiple second music rhythms. By inputting the emotional state feature subsets into the decision tree and obtaining the corresponding sub-decision trees, a matching second music rhythm can be obtained.

[0115] The signal acquisition module 1 includes: an EEG acquisition unit, an fNIRS acquisition unit, and a synchronization acquisition unit.

[0116] The EEG acquisition unit is used to collect brain wave data.

[0117] The fNIRS acquisition unit is used to collect near-infrared light data.

[0118] The synchronous acquisition unit is electrically connected to the EEG acquisition unit and the fNIRS acquisition unit respectively, and is used to synchronously acquire brain wave data and near-infrared light data to obtain synchronized brain wave data and near-infrared light data.

[0119] In one example, the signal acquisition module 1 has eight electrodes as input points for specific physiological parameters. The interfaces of the eight electrodes are of the same size, allowing for free removal of the electrodes. Necessary light shielding is also performed to prevent interference with the electrodes. The EEG acquisition unit has four electrodes for collecting brainwave data, and the fNIRS acquisition unit has four electrodes for collecting near-infrared light data. The synchronization acquisition unit synchronizes the EEG and fNIRS acquisition units. The signal acquisition module 1 is portable and can move freely with the user. It has electromagnetic shielding, enabling data acquisition in complex environments. The fNIRS acquisition unit utilizes a three-wavelength light source and an avalanche diode with amplification to detect weak near-infrared light signals. This highly sensitive device can be used in conjunction with the EEG acquisition unit. The sampling frequency of the fNIRS acquisition unit is generally 17-100Hz, while the sampling frequency of the EEG acquisition unit is generally 1000Hz. Therefore, the synchronization accuracy between the two depends on the sampling frequency of the fNIRS acquisition unit, with an accuracy of up to 10ms.

[0120] Exemplarily, the user's brain wave frequency is lower when the user is resting statically, and the user's brain wave frequency is higher when the user is exercising. The frequency of the user's brain waves during physiological activities is approximately in the range of 0.5-50Hz, and the amplitude is approximately 1-200μV. According to the scene mode preset by the user, such as sleep mode, exercise mode, rest mode, meditation mode, etc., the device automatically matches the main brain wave data (frequency, amplitude, distribution) of the corresponding scene mode. Among them, alpha waves (alpha waves, 8-12Hz) represent rest mode, beta waves (beta waves, 12-40Hz) represent exercise mode, gamma waves (gamma waves, 40-100Hz) represent meditation mode, delta waves (delta waves, 0-4Hz) represent sleep mode, and theta waves (theta waves, 4-8Hz) represent deep sleep mode.

[0121] The data processing module 2 includes: a filtering unit, an artifact recognition unit, and a baseline calibration unit.

[0122] The filtering unit is connected to the synchronization acquisition unit, and is used to filter the synchronized brain wave data and near-infrared light data to obtain filtered brain wave data and near-infrared light data.

[0123] The artifact recognition unit is connected to the filtering unit, and the artifact recognition unit is used to: identify and delete data with voltage values ​​greater than a first threshold in the filtered brain wave data, and identify and delete data with fluctuation deviation values ​​greater than a second threshold in the near-infrared light data, to obtain brain wave data and near-infrared light data after artifact recognition.

[0124] The baseline calibration unit is connected to the artifact recognition unit, and is used to perform a third threshold calibration on the brain wave data and near-infrared light data after artifact recognition in terms of duration to obtain calibrated brain wave data and near-infrared light data.

[0125] In one example, the filtering unit filters out brain wave data and near infrared light data outside of 0.1-30 Hz. The artifact identification unit identifies artifacts with voltage values ​​greater than 300 μV. The baseline calibration unit performs -100 ms baseline calibration.

[0126] To achieve the above objectives, the present invention further provides the following solutions:

[0127] A recommendation system based on EEG and fNIRS, including:

[0128] The first signal acquisition module 6 is used to collect the user's specific physiological parameters; the specific physiological parameters include: brain wave data and near-infrared light data.

[0129] In one example, for the detailed description of the first signal acquisition module 6 , please refer to the description of the signal acquisition module 1 above, which will not be repeated here.

[0130] Cloud server 7 is used for:

[0131] The specific physiological parameters are preprocessed to form preprocessed specific physiological parameters.

[0132] Individual situational features are extracted from the pre-processed specific physiological parameters to obtain behavioral activity features and emotional state features.

[0133] The stored music rhythms are identified and matched through the feature network to obtain the first music rhythm that matches the emotional state characteristics.

[0134] The first music rhythm is identified and matched through a decision tree to obtain a second music rhythm that matches the behavioral activity characteristics.

[0135] In one example, for the detailed description of the cloud server 7 , please refer to the description of the data processing module 2 , the feature analysis module 3 , and the music synthesis module 4 above, which will not be repeated here.

[0136] In addition to storing a large number of user-specific physiological parameters, the cloud server 7 also has an important feature: data security. Information related to the user's life is information that requires privacy protection, and it is safer to store it in the cloud server 7.

[0137] The cloud server 7 can supplement and modify the feature network and decision tree model based on the specific physiological parameters of a large number of users. The cloud server 7 can also be connected to the Internet of Things through a communication module to create a multi-faceted data exchange system.

[0138] The first personalized recommendation module 8 is used to play the second music rhythm to the user.

[0139] In one example, for a detailed description of the first personalized recommendation module 8 , please refer to the description of the personalized recommendation module 5 above, which will not be repeated here.

[0140] The specific process of building a feature network is as follows:

[0141] Collect and analyze the behavioral activity characteristics and emotional state characteristics of multiple users.

[0142] The behavioral activity features of multiple users are grouped into a behavioral activity feature set; a behavioral activity feature set includes multiple different behavioral activity feature subsets.

[0143] The emotional state features of multiple users are combined into an emotional state feature set; an emotional state feature set includes multiple different emotional state feature subsets.

[0144] The stored music rhythms are classified to obtain a plurality of first music rhythms; the plurality of first music rhythms are matched with a plurality of different emotional state feature subsets.

[0145] The first music rhythm is classified to obtain a plurality of second music rhythms; the plurality of second music rhythms are matched with a plurality of different behavioral activity feature subsets.

[0146] In an example, the process of constructing a feature network is described above and will not be described in detail here.

[0147] Identifying and matching the first music rhythm through the decision tree specifically includes:

[0148] The decision tree includes multiple sub-decision trees; the multiple sub-decision trees correspond one-to-one to multiple different emotional state feature subsets; and the multiple different emotional state feature subsets match multiple second music rhythms.

[0149] Multiple subsets of different emotional state features are input into the decision tree to obtain corresponding sub-decision trees.

[0150] The corresponding sub-decision tree decides the matching second music rhythm.

[0151] In one example, the identification and matching of the first music rhythm by using the decision tree is described above and will not be described in detail here.

[0152] The first signal acquisition module 6 includes:

[0153] The EEG acquisition unit is used to collect brain wave data.

[0154] The fNIRS acquisition unit is used to collect near-infrared light data.

[0155] The synchronous acquisition unit is connected to the EEG acquisition unit and the fNIRS acquisition unit respectively. The synchronous acquisition unit is used to synchronously acquire brain wave data and near-infrared light data to obtain synchronized brain wave data and near-infrared light data.

[0156] In one example, for the detailed description of the first signal acquisition module 6 , please refer to the description of the signal acquisition module 1 above, which will not be repeated here.

[0157] The cloud server 7 includes:

[0158] The filtering unit is connected to the synchronization acquisition unit, and is used to filter the synchronized brain wave data and near-infrared light data to obtain filtered brain wave data and near-infrared light data.

[0159] The artifact recognition unit is connected to the filtering unit, and the artifact recognition unit is used to: identify and delete data with voltage values ​​greater than a first threshold in the filtered brain wave data, and identify and delete data with fluctuation deviation values ​​greater than a second threshold in the near-infrared light data, to obtain brain wave data and near-infrared light data after artifact recognition.

[0160] The baseline calibration unit is connected to the artifact recognition unit, and is used to perform a third threshold calibration on the brain wave data and near-infrared light data after artifact recognition in terms of duration to obtain calibrated brain wave data and near-infrared light data.

[0161] In one example, for the detailed description of the cloud server 7 , please refer to the description of the data processing module 2 , the feature analysis module 3 , and the music synthesis module 4 above, which will not be repeated here.

[0162] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0163] This document uses specific examples to illustrate the principles and implementation methods of the embodiments of the present invention. The description of the above embodiments is only intended to help understand the methods and core concepts of the embodiments of the present invention. At the same time, for those skilled in the art, based on the concepts of the embodiments of the present invention, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the embodiments of the present invention.

Claims

1. A wearable device based on EEG and fNIRS, characterized in that: include: A signal acquisition module, used to collect the user's specific physiological parameters; The specific physiological parameters include: brain wave data and near-infrared light data; A data processing module, configured to pre-process the specific physiological parameters and generate pre-processed specific physiological parameters; A feature analysis module is used to extract individual scenario features from the pre-processed specific physiological parameters to obtain behavioral activity features and emotional state features; Temperament synthesis module, used for: Identify and match the stored music rhythms through a feature network to obtain a first music rhythm that matches the emotional state feature; Identify and match the first music rhythm through a decision tree to obtain a second music rhythm that matches the behavioral activity characteristics; A personalized recommendation module is used to play the second music rhythm to the user.

2. The wearable device based on EEG and fNIRS according to claim 1, characterized in that The process of constructing the feature network is specifically as follows: Collect and analyze the behavioral activity characteristics and emotional state characteristics of multiple users; Combining the behavioral activity features of multiple users into a behavioral activity feature set; one behavioral activity feature set includes multiple different behavioral activity feature subsets; Combining the emotional state characteristics of multiple users into an emotional state characteristic set; one of the emotional state characteristic sets includes multiple different emotional state characteristic subsets; Classifying the stored music rhythms to obtain a plurality of first music rhythms; matching the plurality of first music rhythms with the plurality of different emotional state feature subsets; The first music rhythm is classified to obtain a plurality of second music rhythms; the plurality of second music rhythms are matched with the plurality of different behavioral activity feature subsets.

3. The wearable device based on EEG and fNIRS according to claim 2, characterized in that The identifying and matching the first music rhythm by using a decision tree specifically includes: The decision tree includes a plurality of sub-decision trees; the plurality of sub-decision trees correspond one-to-one to the plurality of different emotional state feature subsets; the plurality of different emotional state feature subsets match the plurality of second music rhythms; Inputting the plurality of different emotional state feature subsets into the decision tree to obtain corresponding sub-decision trees; The corresponding sub-decision tree decides a matching second music rhythm.

4. The wearable device based on EEG and fNIRS according to claim 1, characterized in that The signal acquisition module includes: An EEG acquisition unit, configured to acquire the brain wave data; fNIRS acquisition unit, used for acquiring the near-infrared light data; A synchronous acquisition unit is connected to the EEG acquisition unit and the fNIRS acquisition unit respectively, and is used to synchronously acquire the brain wave data and the near-infrared light data to obtain synchronized brain wave data and near-infrared light data.

5. The wearable device based on EEG and fNIRS according to claim 4, characterized in that The data processing module includes: a filtering unit connected to the synchronization acquisition unit, and configured to filter the synchronized brain wave data and near-infrared light data to obtain filtered brain wave data and near-infrared light data; an artifact recognition unit connected to the filtering unit, configured to: identify and delete data having a voltage value greater than a first threshold value in the filtered EEG data, and identify and delete data having a fluctuation deviation value greater than a second threshold value in the near-infrared light data, thereby obtaining artifact-recognized EEG data and near-infrared light data; The baseline calibration unit is connected to the artifact recognition unit and is used to perform a third threshold calibration on the brain wave data and near-infrared light data after artifact recognition in terms of duration to obtain calibrated brain wave data and near-infrared light data.

6. A recommendation system based on EEG and fNIRS, characterized in that: include: A first signal acquisition module is used to collect specific physiological parameters of the user; The specific physiological parameters include: brain wave data and near-infrared light data; Cloud servers for: Preprocessing the specific physiological parameters to form preprocessed specific physiological parameters; Extracting individual situational features from the pre-processed specific physiological parameters to obtain behavioral activity features and emotional state features; Identify and match the stored music rhythms through a feature network to obtain a first music rhythm that matches the emotional state feature; Identify and match the first music rhythm through a decision tree to obtain a second music rhythm that matches the behavioral activity characteristics; The first personalized recommendation module is used to play the second music rhythm to the user.

7. The EEG and fNIRS-based recommendation system according to claim 6, characterized in that The process of constructing the feature network is specifically as follows: Collect and analyze the behavioral activity characteristics and emotional state characteristics of multiple users; Combining the behavioral activity features of multiple users into a behavioral activity feature set; one behavioral activity feature set includes multiple different behavioral activity feature subsets; Combining the emotional state characteristics of multiple users into an emotional state characteristic set; one of the emotional state characteristic sets includes multiple different emotional state characteristic subsets; Classifying the stored music rhythms to obtain a plurality of first music rhythms; matching the plurality of first music rhythms with the plurality of different emotional state feature subsets; The first music rhythm is classified to obtain a plurality of second music rhythms; the plurality of second music rhythms are matched with the plurality of different behavioral activity feature subsets.

8. The EEG and fNIRS-based recommendation system according to claim 7, characterized in that The identifying and matching the first music rhythm by using a decision tree specifically includes: The decision tree includes a plurality of sub-decision trees; the plurality of sub-decision trees correspond one-to-one to the plurality of different emotional state feature subsets; the plurality of different emotional state feature subsets match the plurality of second music rhythms; Inputting the plurality of different emotional state feature subsets into the decision tree to obtain corresponding sub-decision trees; The corresponding sub-decision tree decides a matching second music rhythm.

9. The EEG and fNIRS-based recommendation system according to claim 6, characterized in that The first signal acquisition module includes: An EEG acquisition unit, configured to acquire the brain wave data; fNIRS acquisition unit, used for acquiring the near-infrared light data; A synchronous acquisition unit is connected to the EEG acquisition unit and the fNIRS acquisition unit respectively, and is used to synchronously acquire the brain wave data and the near-infrared light data to obtain synchronized brain wave data and near-infrared light data.

10. The EEG and fNIRS-based recommendation system according to claim 9, characterized in that The cloud server includes: a filtering unit connected to the synchronization acquisition unit, and configured to filter the synchronized brain wave data and near-infrared light data to obtain filtered brain wave data and near-infrared light data; an artifact recognition unit connected to the filtering unit, configured to: identify and delete data having a voltage value greater than a first threshold value in the filtered EEG data, and identify and delete data having a fluctuation deviation value greater than a second threshold value in the near-infrared light data, thereby obtaining artifact-recognized EEG data and near-infrared light data; The baseline calibration unit is connected to the artifact recognition unit and is used to perform a third threshold calibration on the brain wave data and near-infrared light data after artifact recognition in terms of duration to obtain calibrated brain wave data and near-infrared light data.

Images