Olfactory fingerprint construction and olfactory function evaluation method and device based on brain-computer interface
By constructing an olfactory stimulation molecular library and physiological fingerprint through brain-computer interface, the problems of subjectivity and individual differences in traditional olfactory assessment methods have been solved, enabling accurate assessment and early diagnosis of olfactory function.
Patent Information
- Application Number
- CN202510994974.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-18
- Publication Date
- 2025-11-21
AI Technical Summary
Traditional methods for assessing olfactory function rely on the subject’s subjective reports, which are difficult to reflect the encoding, processing and higher cognitive functions of olfactory information in the human brain. They cannot distinguish between damage to the peripheral olfactory system and abnormal processing in the central nervous system. Furthermore, they cannot take into account individualized physiological and psychological characteristics, cannot provide real-time neurophysiological signals, and are difficult to capture instantaneous changes and abnormal patterns in olfactory information processing.
By acquiring the brain's response to olfactory stimuli through a brain-computer interface, an olfactory stimulus molecular library is constructed, background information of olfactory stimulus sources is recorded, and brain physiological activity signals of the test subjects are collected simultaneously. Time-frequency domain analysis is performed to construct an olfactory physiological fingerprint, which is then input into a trained olfactory function assessment model to predict olfactory sensitivity, discrimination ability, memory retrieval ability, and emotional response intensity.
It enables more accurate olfactory function assessment, reflecting individual neural activity patterns under different olfactory stimuli, improving the accuracy and personalization of the assessment, capturing instantaneous changes and abnormal patterns in olfactory information processing, and is suitable for the early detection and diagnosis of olfactory disorders.
Smart Images
Figure CN120994053A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of neuroscience and artificial intelligence, specifically to a method and device for constructing olfactory fingerprints and evaluating olfactory function based on a brain-computer interface, as well as computer equipment. Background Technology
[0002] Olfactory dysfunction, such as decreased sense of smell, loss of smell, or olfactory inversion, affects the quality of life of a large number of people. It is often an early symptom or complication of many diseases, such as neurological and mental illnesses (e.g., Alzheimer's disease, Parkinson's disease, anxiety, depression), viral infections, traumatic brain injury, or chronic rhinitis, and is closely related to the occurrence and development of these diseases.
[0003] Traditional methods for assessing olfactory function primarily rely on psychophysical tests, such as olfactory recognition threshold tests, odor recognition tests, olfactory discrimination tests, and odor memory tests. These tests assess olfactory ability by having subjects smell different odors and report their perception or recognition results. First, these methods are dependent on factors such as the subject's subjective report, comprehension ability, attention level, and the testing environment. For example, subjects may experience fatigue, anxiety, or deliberately conceal their true feelings, affecting the accuracy of test results and making reliable comparisons and quantification difficult between different test subjects or between different time points for the same test subject. For children, patients with cognitive impairments, or patients with impaired consciousness, these methods are difficult to implement or obtain effective data from. Second, they cannot directly reflect the physiological processes of encoding, processing, and higher cognitive functions (such as memory and emotion) of olfactory information in the human brain. Olfactory perception is a complex, multi-layered neural activity process involving the coordinated actions of multiple brain regions, from olfactory receptors to the olfactory bulb, piriform cortex, amygdala, hippocampus, and even the prefrontal cortex. Behavioral reports alone cannot fully reveal the neural mechanisms behind olfactory disorders, nor can they distinguish whether the dysfunction is caused by damage to the peripheral olfactory system or by abnormal processing in the central nervous system. Finally, individual differences in olfaction are not only reflected in olfactory sensitivity but also in the perception, preference, and memory of specific odors. Traditional testing methods typically use standardized odor databases, which fail to adequately consider these individualized physiological and psychological characteristics, thus limiting the accuracy and personalization of assessments. For example, different people may perceive the same odor with different intensities, levels of pleasure, and even generate drastically different associations and memories. Furthermore, the neural activity of the brain is dynamic during the presentation of olfactory stimuli, and traditional methods cannot provide real-time neurophysiological signals, making it difficult to capture instantaneous changes and abnormal patterns in olfactory information processing. For example, the brain's reaction speed to different odors, changes in oscillation patterns, and the coupling strength of specific brain regions may all carry important information about olfactory function.
[0004] To address the aforementioned issues, this invention proposes a brain-computer interface-based method for constructing olfactory fingerprints and assessing olfactory function. By acquiring the brain's response to olfactory stimuli through a brain-computer interface, the method reflects an individual's neural activity patterns under different olfactory stimuli, thereby more accurately assessing olfactory sensitivity, discrimination ability, memory retrieval ability, and emotional response intensity. Summary of the Invention
[0005] The technical problem to be solved by the present invention is to provide a method, device and computer equipment for constructing olfactory fingerprints and evaluating olfactory function based on brain-computer interface, which addresses the shortcomings of the prior art.
[0006] According to one aspect of the present invention, a method for constructing olfactory fingerprints and evaluating olfactory function based on a brain-computer interface is provided, comprising:
[0007] An olfactory stimulus molecular library was constructed based on the individualized information of the test subjects; wherein, the individualized information includes the olfactory receptor genotype data and olfactory preference / dislike memories of the test subjects; the olfactory stimulus molecular library includes pleasantness, intensity, familiarity, complexity features and corresponding olfactory stimulus spectra;
[0008] An olfactory stimulus of preset concentration, flow rate, and duration is provided into the nasal cavity of the subject, and background information of the olfactory stimulus is recorded; wherein, the background information includes the semantic label of the odor source, the intensity of emotional evokedness, the intensity of memory association, or the subject's association report with the odor source;
[0009] During the presentation of olfactory stimulation and within a preset time after the stimulation, the brain physiological activity signals of the subject are synchronously collected through the brain-computer interface cap. The brain-computer interface cap includes a high-density electrode array, an eye-tracking sensor, an electromyography sensor, and a near-infrared spectroscopy sensor array.
[0010] Based on the olfactory stimulation molecule library, the brain physiological activity signals of the test subjects were analyzed in the time and frequency domain. Based on the extracted microstate sequences, phase-amplitude coupling strength and high-frequency oscillation power distribution related to the olfactory information encoding, an olfactory physiological fingerprint was constructed.
[0011] The olfactory physiological fingerprint is input into a trained domain-adaptive olfactory function assessment model to predict the subject's current olfactory sensitivity, discrimination ability, memory retrieval ability, and emotional response intensity.
[0012] In one alternative approach, when extracting the high-frequency oscillation power distribution related to olfactory information encoding based on the subject's brain physiological activity signals, the high-frequency oscillation power at each time point is calculated; wherein, the formula for calculating the high-frequency oscillation power is:
[0013]
[0014] in, and It is the preset time range during and after the presentation of olfactory stimuli; and To set the frequency range; It is a probability weighting factor; For frequency The time-frequency distribution function at time point t; Let be the rate of change of power at time t. , For time intervals, For the signal at time point t within the oscillation range [ Total power within; , The number of window functions. For the first Brain physiological activity signals at each sampling point is the window function at time point t.
[0015] In an alternative approach, the method further includes:
[0016] The dynamic complexity index of a microstate is calculated based on the probability of its occurrence at a given time point and the transition probability of that microstate. The formula for calculating the dynamic complexity index is as follows:
[0017]
[0018] in, The number of microstates; It is a sequence of microstates; To start from microstate The probability of transitioning to microstate j; For observation time; For the microstate at time t The probability of occurrence.
[0019] In an alternative approach, the method further includes:
[0020] The low-frequency phase signal is divided into multiple phase intervals, and the average value of the high-frequency amplitude signal in each phase interval is calculated.
[0021] Based on the influence of different frequency combinations on the encoding of olfactory information, the coupling index of all possible low-frequency / high-frequency combinations is calculated;
[0022] The phase / amplitude coupling strength characteristics are improved based on weighting factors and coupling exponents for all possible low / high frequency combinations; where the coupling strength characteristics are:
[0023]
[0024] in, To measure the weighting factor of the combination of low frequency i and high frequency j in olfactory perception; Let be the coupling index of the frequency combination (i,j). N is the total number of phase intervals into which the low-frequency phase signal is divided. For a phase signal of a specific low frequency i to fall within the k-th phase interval The average value of the amplitude signal at the corresponding high-frequency j. For a phase signal of a specific low frequency i to fall within the l-th phase interval The average value of the amplitude signal at the corresponding high-frequency j.
[0025] , ;
[0026] For falling within the phase interval Number of time points It is the phase signal of low frequency i at time t.
[0027] In an alternative approach, constructing the olfactory stimulus molecular library based on the individualized information of the test subject further includes:
[0028] The complete olfactory receptor gene sequence of the subject was obtained using gene sequencing methods.
[0029] The olfactory receptor gene sequences were compared and analyzed to identify key gene loci and mutation types.
[0030] The olfactory stimulation molecule library is constructed by associating the gene loci, mutation types, and memory information.
[0031] In one alternative approach, the domain-adaptive olfactory function assessment model includes a physiological fingerprint encoder, a background information encoder, a cross-modal fusion and attention layer, a domain-invariant feature learning layer, and an olfactory function prediction head.
[0032] The physiological fingerprint encoder includes a micro-state sequence processing module, a phase-amplitude coupling processing module, and a phase-amplitude coupling processing module.
[0033] The background information encoder includes a text information encoding module and a numerical information encoding module;
[0034] The cross-modal fusion and attention layer dynamically control the contribution ratio of different modal information through a gated multimodal fusion unit;
[0035] The domain-invariant feature learning layer includes a feature extractor, a domain discriminator, and a joint optimization objective;
[0036] The olfactory function prediction head includes a multi-task predictor composed of multiple independent or shared underlying fully connected neural network branches, with each neural network branch used to predict an olfactory function index.
[0037] In one alternative approach, the microstate sequence processing module employs a Bi-GRU bidirectional gated cyclic unit network layer, wherein the GRU units in the Bi-GRU bidirectional gated cyclic unit network layer include update gates and reset gates;
[0038] The phase-amplitude coupling processing module includes a one-dimensional convolutional neural network layer and a two-dimensional convolutional neural network layer;
[0039] The gated multimodal fusion unit adopts a dynamic routing attention mechanism. The dynamic routing mechanism allocates the initial fusion features to different capsules by iteratively updating the routing weights. Each capsule represents an olfactory feature mode. The outputs of all capsules are concatenated to obtain cross-modal fusion features.
[0040] The neighborhood discriminator uses a GRL gradient inversion layer to connect the feature extractor and the neighborhood discriminator; wherein, the GRL gradient inversion layer keeps the input unchanged during forward propagation and inverts the gradient during backward propagation;
[0041] The olfactory function prediction head uses different activation functions for different olfactory function indicators; a linear activation function is used for olfactory sensitivity and discrimination ability, and a softmax activation function is used for emotional response intensity level.
[0042] In an alternative approach, the time-frequency domain analysis of the subject's brain physiological activity signals based on the olfactory stimulation molecule library further includes:
[0043] Based on the characteristics of pleasure, intensity, familiarity, and complexity of the olfactory stimulus molecular library and the olfactory stimulus spectrum, the olfactory stimuli were grouped, and the brain physiological activity signals corresponding to each group of stimuli were decomposed in time and frequency.
[0044] In each olfactory stimulation group, the number of clusters or convergence criteria are guided by the complexity characteristics of each stimulus in the olfactory stimulation molecule library to obtain a prototype map of brain microstates under that stimulation group; the brain topography map at each time point is matched with the obtained prototype map of microstates, and the duration, frequency of occurrence and transition probability between microstates under different stimulation groups are calculated to obtain a microstate sequence.
[0045] Before calculating the average value of the high-frequency amplitude signal in each phase interval, the low-frequency phase signal and the high-frequency amplitude signal are extracted by Hilbert transform.
[0046] Before calculating the high-frequency oscillation power at each time point, the power values of the high-gamma band under different olfactory stimulation groups are calculated and the distribution of the high-density electrode arrays in different brain-computer interface caps is analyzed to obtain the high-frequency oscillation power distribution related to the emotional induction intensity and memory association intensity of each stimulus in the olfactory stimulation molecular library.
[0047] According to another aspect of the present invention, a brain-computer interface-based olfactory fingerprint construction and olfactory function assessment device is provided, comprising:
[0048] The olfactory stimulus molecular library construction module is used to construct an olfactory stimulus molecular library based on the individualized information of the test subjects; wherein, the individualized information includes the olfactory receptor genotype data and olfactory preference / aversion memory of the test subjects; the olfactory stimulus molecular library includes pleasantness, intensity, familiarity, complexity features and corresponding olfactory stimulus spectra;
[0049] The olfactory stimulus presentation and background information recording module is used to provide the subject with an olfactory stimulus source of preset concentration, flow rate and duration in the nasal cavity, and record the background information of the olfactory stimulus source; wherein, the background information includes the semantic label of the odor source, the intensity of emotional induction, the intensity of memory association or the subject's association report with the odor source;
[0050] The brain physiological activity signal acquisition module is used to synchronously acquire the brain physiological activity signals of the subject through a brain-computer interface cap during the presentation of olfactory stimulation and within a preset time after the presentation of stimulation. The brain-computer interface cap includes a high-density electrode array, an eye-tracking sensor, an electromyography sensor, and a near-infrared spectroscopy sensor array.
[0051] The olfactory physiological fingerprint construction module is used to perform time-frequency domain analysis on the brain physiological activity signals of the subject based on the olfactory stimulation molecule library, and to construct the olfactory physiological fingerprint based on the extracted microstate sequence, phase-amplitude coupling strength and high-frequency oscillation power distribution related to the encoding of olfactory information.
[0052] The olfactory function assessment module is used to input the olfactory physiological fingerprint into a trained domain-adaptive olfactory function assessment model to predict the subject's current olfactory sensitivity, discrimination ability, memory retrieval ability, and emotional response intensity.
[0053] According to another aspect of the present invention, a computer device is provided, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other through the communication bus;
[0054] The memory is used to store at least one executable instruction, which causes the processor to perform the operations corresponding to the above-described brain-computer interface-based olfactory fingerprint construction and olfactory function assessment method.
[0055] According to the scheme provided by the present invention, an olfactory stimulus molecular library is constructed based on the individualized information of the test subject; wherein, the individualized information includes the test subject's olfactory receptor genotype data and olfactory preference / aversion memories; the olfactory stimulus molecular library includes pleasantness, intensity, familiarity, complexity characteristics and corresponding olfactory stimulus spectra; olfactory stimuli of preset concentration, flow rate and duration are provided to the test subject's nasal cavity, and background information of the olfactory stimuli is recorded; wherein, the background information includes the semantic label of the odor source, the intensity of emotional induction, the intensity of memory association, or the test subject's association report with the odor source; preset concentrations, flow rates and durations of olfactory stimuli are provided during and after the presentation of olfactory stimuli. Within a given timeframe, brain physiological activity signals of the subject are synchronously acquired via a brain-computer interface cap, which includes a high-density electrode array, an eye-tracking sensor, an electromyography (EMG) sensor, and a near-infrared spectroscopy sensor array. Time-frequency domain analysis is performed on the subject's brain physiological activity signals based on an olfactory stimulus molecular library. An olfactory physiological fingerprint is constructed based on the extracted microstate sequences related to olfactory information encoding, phase-amplitude coupling strength, and high-frequency oscillation power distribution. This olfactory physiological fingerprint is then input into a trained domain-adaptive olfactory function assessment model to predict the subject's current olfactory sensitivity, discrimination ability, memory retrieval ability, and emotional response intensity. This invention acquires the brain's response to olfactory stimuli through a brain-computer interface, reflecting an individual's neural activity patterns under different olfactory stimuli, thereby more accurately assessing olfactory sensitivity, discrimination ability, memory retrieval ability, and emotional response intensity.
[0056] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description
[0057] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:
[0058] Figure 1 This illustration shows a flowchart of the brain-computer interface-based olfactory fingerprint construction and olfactory function assessment method according to an embodiment of the present invention. Figure 1 ;
[0059] Figure 2 This illustration shows a flowchart of the brain-computer interface-based olfactory fingerprint construction and olfactory function assessment method according to an embodiment of the present invention. Figure 2 ;
[0060] Figure 3 This invention illustrates a schematic diagram of the framework of a brain-computer interface-based olfactory fingerprint construction and olfactory function assessment device according to an embodiment of the present invention.
[0061] Figure 4 A schematic diagram of the structure of a computer device according to an embodiment of the present invention is shown. Detailed Implementation
[0062] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
[0063] Figure 1 This illustration shows a flowchart of the brain-computer interface-based olfactory fingerprint construction and olfactory function assessment method according to an embodiment of the present invention. Figure 1 , Figure 1 This illustration shows a flowchart of the brain-computer interface-based olfactory fingerprint construction and olfactory function assessment method according to an embodiment of the present invention. Figure 2 Specifically, such as Figure 1 , Figure 2 As shown, it includes the following steps:
[0064] Step S101: Construct an olfactory stimulus molecular library based on the individualized information of the test subject; wherein, the individualized information includes the olfactory receptor genotype data and olfactory preference / dislike memory of the test subject; the olfactory stimulus molecular library includes pleasantness, intensity, familiarity, complexity features and corresponding olfactory stimulus spectra.
[0065] In this embodiment, traditional olfactory assessments often employ standardized odor libraries, ignoring individual differences in olfactory perception. This embodiment, by introducing olfactory receptor genotype data and olfactory preference / aversion memories as personalized information, constructs an olfactory stimulus molecular library that highly matches the test subject's physiological basis and psychological background, more accurately reflecting the individual's olfactory ability and characteristics. Since different individuals may perceive the same odor drastically, influenced not only by olfactory receptor genes but also by personal experiences and cultural background, combining genotype and memory allows for the selection or design of stimuli that are more physiologically or psychologically significant for a specific individual. This avoids situations where some stimuli may be too weak or too strong for some people, leading to ineffective assessment of their true olfactory ability, thus improving assessment sensitivity and diagnostic potential. For example, a person naturally insensitive to a certain odor may not effectively elicit a brain response if tested with that odor; however, if a odor sensitive to their genotype is selected, their weak olfactory signals can be better captured, thus helping to detect potential olfactory dysfunctions (such as olfactory decline caused by early-stage Parkinson's disease, Alzheimer's disease, and other neurodegenerative diseases) earlier and more accurately. The olfactory stimulus molecular library not only contains specific stimulus molecules, but also includes pleasure, intensity, familiarity, complexity, and olfactory stimulus spectrum. The analysis of brain physiological activity can provide a deeper understanding of the complexity of olfactory information encoding, rather than just recognizing a certain odor, thereby constructing a more comprehensive olfactory physiological fingerprint.
[0066] For example, Subject A possesses a specific SNP (such as rsxxxx) on the OR7G2 olfactory receptor gene, which is known in the population to be significantly associated with the perception of "cilantro smell" (especially the bitter / soapy smell of its aldehyde compounds). Subject A also exhibits a wild-type on the OR2M7 olfactory receptor gene, which is associated with sensitivity to β-ionone. Subject A expressed a strong aversion to the smell of cilantro, describing it as "like soapy water," and has disliked it since childhood. This stems from Subject A's childhood memory of jasmine flowers growing in front of their house, leading them to particularly enjoy the scent of jasmine, finding it relaxing and pleasant, and associating it with carefree childhood memories. When presented with a range of odor samples, Subject A rated the pleasantness of "jasmine essence" at 10 (highest) and the memory association strength at 5 (highest); the pleasantness of "cilantro extract" at 1 (lowest) and the memory association strength at 4 (highest); and the pleasantness of "coffee essence" at 5 (medium) and the memory association strength at 2 (lowest). Based on the individualized information above, the following olfactory stimulus molecular library was constructed for subject A (partial example):
[0067] Stimulant molecule 1: trans-2-decenal (one of the main flavor compounds in coriander);
[0068] Genotype association: Highly associated with subject A's OR7G2 genotype, suggesting that subject A is expected to be sensitive to it and tend to perceive it negatively.
[0069] Preference / Aversion Memory: Strong aversion, associated with early memories. Pleasure: Low (expected); Intensity: High (expected); Familiarity: High (familiar since childhood); Complexity: Moderate (single molecule); Olfactory Stimulus Spectrum: Green leaves, aldehydes, slightly soapy / metallic; Stimulant molecule 2: Jasmine ketone (one of the main aroma components of jasmine) or jasmine essence; Genotype Association: No direct and significant genotype association.
[0070] In an alternative approach, constructing the olfactory stimulus molecular library based on the individualized information of the test subject further includes:
[0071] The complete olfactory receptor gene sequence of the subject was obtained using gene sequencing methods.
[0072] The olfactory receptor gene sequences were compared and analyzed to identify key gene loci and mutation types.
[0073] The olfactory stimulation molecule library is constructed by associating the gene loci, mutation types, and memory information.
[0074] In this embodiment, for example, a saliva sample was collected from subject A and whole-exome sequencing was performed. The sequencing results showed that subject A had a SNP mutation in a key olfactory receptor gene, OR2M3. The OR2M3 gene sequence of subject A was compared with a reference sequence, confirming a T mutation. This T mutation in the OR2M3 gene is associated with reduced sensitivity to specific amygdalin-like odors (such as almond or cherry). That is, individuals carrying the T mutation may require higher concentrations of almond odor to perceive it, or have a weaker perception of almond odor. Based on the analysis of the OR2M3 gene, the olfactory stimulus molecular library identified amygdalin (representing almond odor) as one of the key stimuli. Different concentration gradients were designed for subject A based on the gene analysis results, or it was used as a key odor for testing subject A's olfactory threshold. In a questionnaire survey, subject A expressed a strong liking for citrus odors (such as lemon or orange) and these odors always associated them with positive memories of sunshine and vitality. At the same time, they showed extreme aversion to the smell of rotten sulfur compounds, which triggered strong negative emotions. The olfactory stimulus molecular library constructed in this embodiment not only includes stimuli traditionally selected based on odor characteristics and subjective feedback, but also includes specific odors that are highly correlated with the subject's genotype, ensuring a more comprehensive assessment of the subject's olfactory sensitivity, discrimination ability, memory retrieval ability, and emotional response intensity.
[0075] Step S102: Provide the subject with an olfactory stimulus source of preset concentration, flow rate and duration into the nasal cavity, and record the background information of the olfactory stimulus source; wherein, the background information includes the semantic label of the odor source, the intensity of emotional induction, the intensity of memory association or the subject's association report with the odor source.
[0076] In this embodiment, odor concentration, flow rate, and duration directly affect the activation level of olfactory receptors and the intensity of olfactory perception, helping to attribute physiological responses to the specificity of the odor rather than differences in physical stimuli. Semantic labels provide objective odor descriptions, facilitating classification and understanding of the subject's basic cognition. Emotional evoked intensity quantifies the emotional response evoked by the odor (e.g., pleasure, aversion, calmness, excitement). Memory association strength assesses the degree to which the odor evokes specific memories. Association reports provide open-ended, individualized subjective experiences, capturing richer and more nuanced olfactory perceptions.
[0077] For example, assess a 60-year-old subject suspected of having mild cognitive impairment accompanied by decreased olfactory sensitivity and memory loss. Three common odors with different emotional and memory association potentials are selected: Odor A: Rose essence (typically associated with pleasure, romance, and gardens); Odor B: Lemon essence (typically associated with freshness, vitality, and cleanliness); Odor C: Musty odor (typically associated with discomfort, dampness, and decay). All odors are preset to the same concentration (0.1% solution saturated vapor), the same flow rate (4 liters / minute), and the same duration (3 seconds) using an olfactory meter. Each odor is presented 5 times, with each stimulus spaced 20 seconds apart, and pure air is randomly interspersed as a control. After each odor presentation (or after each set of odor presentations), the subject is asked and the question is recorded: "What does odor A remind you of?" (The subject might answer: floral, rose, perfume). "When you smell odor A, do you feel happy or unhappy? How happy? (1-7 points, 1 very unhappy, 7 very happy)." "Does odor A remind you of a specific person, place, or event? How strong? (1-7 points, 1 no memory, 7 strong memory) If so, please describe briefly." "Besides 'rose,' do you have any other associations with odor A? For example, images, sounds, or feelings?" The above background information is used as labels and mapped to the brain physiological activity signals (such as microstate sequences, phase-amplitude coupling, and high-frequency oscillation power distribution) collected from the subject through a brain-computer interface cap when smelling the above odors. If the subject's semantic label is vague, emotional intensity score is low, and memory association is low when smelling rose essence, and the physiological fingerprint shows weak encoding activity for that odor, it suggests a decline in olfactory sensitivity, emotional response, and memory retrieval ability. If the physiological fingerprint encoding of rose essence is normal, but the subjectively reported emotional and memory associations are weak, it suggests that olfactory perception is basically normal, but there are obstacles at the higher cognitive and emotional levels.
[0078] Step S103: During the presentation of olfactory stimulation and within a preset time after the presentation of stimulation, the brain physiological activity signals of the subject are synchronously collected through the brain-computer interface cap. The brain-computer interface cap includes a high-density electrode array, an eye-tracking sensor, an electromyography sensor, and a near-infrared spectroscopy sensor array.
[0079] In this embodiment, a high-density electrode array captures the electrical activity of brain neurons, reflecting the temporal dynamics and spatial distribution of olfactory information encoding, cognitive processing, and emotional responses in the cerebral cortex. An eye-tracking sensor provides additional attention, cognitive load, and emotional cues; combined with EEG data, it can reveal how olfactory stimuli guide visual attention or how visual information, in turn, influences olfactory perception. An electromyography (EMG) sensor monitors muscle activity, particularly capturing subtle movements of facial expression muscles. A near-infrared spectroscopy sensor array measures changes in blood oxygen levels in the cerebral cortex, reflecting local hemodynamic responses caused by neural activity. It provides blood oxygen metabolism information complementary to EEG, especially in deep cortical regions or when EEG signals are affected by artifacts, offering robust indicators of neural activity and providing a more comprehensive physiological context for interpreting EEG signals. The cap incorporates a pre-designed high-density electrode layout and integrates an eye-tracking camera (located in the forehead or below the eye socket), EMG electrodes (attached near the orbicularis oculi and zygomaticus major muscles), and a near-infrared light source and detector array (located above cortical regions related to olfactory perception, such as the frontal and temporal lobes). Before the experiment begins, technicians assist the participants in properly wearing the brain-computer interface cap, ensuring good contact between all sensors and the skin (or eyes). Data signals from all sensors are collected by a small data acquisition unit inside or outside the cap, which is responsible for analog-to-digital conversion, signal amplification, and synchronization timestamp marking.
[0080] Step S104: Perform time-frequency domain analysis on the brain physiological activity signals of the subject based on the olfactory stimulation molecule library, and construct an olfactory physiological fingerprint based on the extracted microstate sequence, phase-amplitude coupling strength and high-frequency oscillation power distribution related to the olfactory information encoding.
[0081] In this embodiment, for example, parameters of the microstate sequence under rose fragrance stimulation (such as the duration, frequency of occurrence, and dynamic complexity index of microstate B), the Theta-Gamma coupling strength PC value, and the high gamma power values and their association strength with emotion / memory in regions such as the medial prefrontal cortex, amygdala, and parahippocampal gyrus are integrated into a multi-dimensional feature vector, which is the subject's personalized olfactory physiological fingerprint of rose fragrance.
[0082] In one alternative approach, when extracting the high-frequency oscillation power distribution related to olfactory information encoding based on the subject's brain physiological activity signals, the high-frequency oscillation power at each time point is calculated; wherein, the formula for calculating the high-frequency oscillation power is:
[0083]
[0084] in, and It is the preset time range during and after the presentation of olfactory stimuli; and To set the frequency range; It is a probability weighting factor; For frequency The time-frequency distribution function at time point t; Let be the rate of change of power at time t. , For time intervals, For the signal at time point t within the oscillation range [ Total power within; , The number of window functions. For the first Brain physiological activity signals at each sampling point is the window function at time point t.
[0085] In this embodiment, high-frequency oscillations, particularly in the gamma band (typically referring to frequencies above 30Hz, or even higher), are considered the neural basis for higher cognitive functions and information processing. By calculating the power of high-frequency oscillations, the synchronous firing intensity of neuronal groups in the brain during olfactory information processing can be quantified, directly reflecting the activity and efficiency of olfactory information encoding. Compared to relying solely on low-frequency oscillations or event-related potentials, high-frequency oscillation power is more sensitive to subtle differences in individual olfactory perception. The power change rate can capture the dynamic trend of power changes in the brain's response to olfactory stimuli, i.e., whether the power is rapidly increasing, decreasing, or stabilizing. The time-frequency distribution function uses a window function to segment the signal, avoiding the limitations of traditional Fourier transform in processing non-stationary signals and better capturing the instantaneous frequency components of the brain signal.
[0086] In an alternative approach, the method further includes:
[0087] The dynamic complexity index of a microstate is calculated based on the probability of its occurrence at a given time point and the transition probability of that microstate. The formula for calculating the dynamic complexity index is as follows:
[0088]
[0089] in, The number of microstates; It is a sequence of microstates; To start from microstate The probability of transitioning to microstate j; For observation time; For the microstate at time t The probability of occurrence.
[0090] In this embodiment, the dynamic complexity index not only considers the probability distribution and transition characteristics of the microstate itself, but also introduces the rate of change of the probability of the microstate occurring over time, representing the dynamic instability and adaptability of the brain when switching from one microstate to another. It can capture the complexity of the brain's functional network reconstruction and state transition when processing olfactory information.
[0091] In an alternative approach, the method further includes:
[0092] The low-frequency phase signal is divided into multiple phase intervals, and the average value of the high-frequency amplitude signal in each phase interval is calculated.
[0093] Based on the influence of different frequency combinations on the encoding of olfactory information, the coupling index of all possible low-frequency / high-frequency combinations is calculated;
[0094] The phase / amplitude coupling strength characteristics are improved based on weighting factors and coupling exponents for all possible low / high frequency combinations; where the coupling strength characteristics are:
[0095]
[0096] in, To measure the weighting factor of the combination of low frequency i and high frequency j in olfactory perception; Let be the coupling index of the frequency combination (i,j). N is the total number of phase intervals into which the low-frequency phase signal is divided. For a phase signal of a specific low frequency i to fall within the k-th phase interval The average value of the amplitude signal at the corresponding high-frequency j. For a phase signal of a specific low frequency i to fall within the l-th phase interval The average value of the amplitude signal at the corresponding high frequency j;
[0097] , ;
[0098] For falling within the phase interval Number of time points It is the phase signal of low frequency i at time t.
[0099] In this embodiment, by considering all possible combinations of low / high frequencies and introducing weighting factors, the influence of the interaction between different frequency oscillations in the brain on the encoding of olfactory information can be captured more comprehensively and precisely, and the complex neural oscillation patterns in the olfactory perception process can be better reflected.
[0100] For example, when subject A smells a specific odor (e.g., lemon), brain activity signals are collected via a brain-computer interface cap. Several low-frequency bands (e.g., Theta waves: 4-8 Hz, Alpha waves: 8-12 Hz) and several high-frequency bands (e.g., low Gamma waves: 30-50 Hz, high Gamma waves: 60-90 Hz) are defined. The phase of the Theta wave (low frequency i) and the amplitude of the low Gamma wave (high frequency j) are extracted. The phase of the Theta wave is divided into 18 intervals, and the average amplitude of the low Gamma wave falling within each phase interval is calculated. If the lemon scent strongly modulates the amplitude of the low Gamma wave, causing it to fluctuate significantly with changes in the Theta wave phase, then the signal is considered.
[0101] In an alternative approach, the time-frequency domain analysis of the subject's brain physiological activity signals based on the olfactory stimulation molecule library further includes:
[0102] Based on the characteristics of pleasure, intensity, familiarity, and complexity of the olfactory stimulus molecular library and the olfactory stimulus spectrum, the olfactory stimuli were grouped, and the brain physiological activity signals corresponding to each group of stimuli were decomposed in time and frequency.
[0103] In each olfactory stimulation group, the number of clusters or convergence criteria are guided by the complexity characteristics of each stimulus in the olfactory stimulation molecule library to obtain a prototype map of brain microstates under that stimulation group; the brain topography map at each time point is matched with the obtained prototype map of microstates, and the duration, frequency of occurrence and transition probability between microstates under different stimulation groups are calculated to obtain a microstate sequence.
[0104] Before calculating the average value of the high-frequency amplitude signal in each phase interval, the low-frequency phase signal and the high-frequency amplitude signal are extracted by Hilbert transform.
[0105] Before calculating the high-frequency oscillation power at each time point, the power values of the high-gamma band under different olfactory stimulation groups are calculated and the distribution of the high-density electrode arrays in different brain-computer interface caps is analyzed to obtain the high-frequency oscillation power distribution related to the emotional induction intensity and memory association intensity of each stimulus in the olfactory stimulation molecular library.
[0106] In this embodiment, not only is high gamma power calculated, but its distribution in the high-density electrode arrays of different brain-computer interface caps is also analyzed. It is correlated with the intensity of emotional evoked emotions and the intensity of memory association, which can better locate the neural activities related to the above functions, thereby gaining a deeper understanding of how the sense of smell triggers emotional and memory responses, and making the fingerprint of high-frequency oscillation power more interpretable and diagnostic.
[0107] Step S105: Input the olfactory physiological fingerprint into the trained domain-adaptive olfactory function assessment model to predict the subject's current olfactory sensitivity, discrimination ability, memory retrieval ability, and emotional response intensity.
[0108] In this embodiment, the domain-adaptive olfactory function assessment model includes a physiological fingerprint encoder, a background information encoder, a cross-modal fusion and attention layer, a domain-invariant feature learning layer, and an olfactory function prediction head;
[0109] The physiological fingerprint encoder includes a micro-state sequence processing module, a phase-amplitude coupling processing module, and a phase-amplitude coupling processing module.
[0110] The background information encoder includes a text information encoding module and a numerical information encoding module;
[0111] The cross-modal fusion and attention layer dynamically control the contribution ratio of different modal information through a gated multimodal fusion unit;
[0112] The domain-invariant feature learning layer includes a feature extractor, a domain discriminator, and a joint optimization objective;
[0113] The olfactory function prediction head includes a multi-task predictor composed of multiple independent or shared underlying fully connected neural network branches, with each neural network branch used to predict an olfactory function index.
[0114] The micro-state sequence processing module adopts a Bi-GRU bidirectional gated cyclic unit network layer, wherein the GRU unit in the Bi-GRU bidirectional gated cyclic unit network layer includes an update gate and a reset gate;
[0115] The phase-amplitude coupling processing module includes a one-dimensional convolutional neural network layer and a two-dimensional convolutional neural network layer;
[0116] The gated multimodal fusion unit adopts a dynamic routing attention mechanism. The dynamic routing mechanism allocates the initial fusion features to different capsules by iteratively updating the routing weights. Each capsule represents an olfactory feature mode. The outputs of all capsules are concatenated to obtain cross-modal fusion features.
[0117] The neighborhood discriminator uses a GRL gradient inversion layer to connect the feature extractor and the neighborhood discriminator; wherein, the GRL gradient inversion layer keeps the input unchanged during forward propagation and inverts the gradient during backward propagation;
[0118] The olfactory function prediction head uses different activation functions for different olfactory function indicators; a linear activation function is used for olfactory sensitivity and discrimination ability, and a softmax activation function is used for emotional response intensity level.
[0119] According to the scheme provided by the present invention, an olfactory stimulus molecular library is constructed based on the individualized information of the test subject; wherein, the individualized information includes the test subject's olfactory receptor genotype data and olfactory preference / aversion memories; the olfactory stimulus molecular library includes pleasantness, intensity, familiarity, complexity characteristics and corresponding olfactory stimulus spectra; olfactory stimuli of preset concentration, flow rate and duration are provided to the test subject's nasal cavity, and background information of the olfactory stimuli is recorded; wherein, the background information includes the semantic label of the odor source, the intensity of emotional induction, the intensity of memory association, or the test subject's association report with the odor source; preset concentrations, flow rates and durations of olfactory stimuli are provided during and after the presentation of olfactory stimuli. Within a given timeframe, brain physiological activity signals of the subject are synchronously acquired via a brain-computer interface cap, which includes a high-density electrode array, an eye-tracking sensor, an electromyography (EMG) sensor, and a near-infrared spectroscopy sensor array. Time-frequency domain analysis is performed on the subject's brain physiological activity signals based on an olfactory stimulus molecular library. An olfactory physiological fingerprint is constructed based on the extracted microstate sequences related to olfactory information encoding, phase-amplitude coupling strength, and high-frequency oscillation power distribution. This olfactory physiological fingerprint is then input into a trained domain-adaptive olfactory function assessment model to predict the subject's current olfactory sensitivity, discrimination ability, memory retrieval ability, and emotional response intensity. This invention acquires the brain's response to olfactory stimuli through a brain-computer interface, reflecting an individual's neural activity patterns under different olfactory stimuli, thereby more accurately assessing olfactory sensitivity, discrimination ability, memory retrieval ability, and emotional response intensity.
[0120] Figure 3 A schematic diagram of the framework of a brain-computer interface-based olfactory fingerprint construction and olfactory function assessment device according to an embodiment of the present invention is shown. The brain-computer interface-based olfactory fingerprint construction and olfactory function assessment device includes:
[0121] The olfactory stimulation molecular library construction module 310 is used to construct an olfactory stimulation molecular library based on the individualized information of the test subject; wherein, the individualized information includes the olfactory receptor genotype data and olfactory preference / dislike memory of the test subject; the olfactory stimulation molecular library includes pleasantness, intensity, familiarity, complexity features and corresponding olfactory stimulation spectra;
[0122] The olfactory stimulus presentation and background information recording module 320 is used to provide the subject with an olfactory stimulus source of preset concentration, flow rate and duration in the nasal cavity, and record the background information of the olfactory stimulus source; wherein, the background information includes the semantic label of the odor source, the intensity of emotional induction, the intensity of memory association or the subject's association report with the odor source;
[0123] The brain physiological activity signal acquisition module 330 is used to synchronously acquire the brain physiological activity signals of the subject through the brain-computer interface cap during the presentation of olfactory stimulation and within a preset time after the presentation of stimulation. The brain-computer interface cap includes a high-density electrode array, an eye-tracking sensor, an electromyography sensor, and a near-infrared spectroscopy sensor array.
[0124] The olfactory physiological fingerprint construction module 340 is used to perform time-frequency domain analysis on the brain physiological activity signals of the subject based on the olfactory stimulation molecule library, and construct an olfactory physiological fingerprint based on the extracted microstate sequence, phase-amplitude coupling strength and high-frequency oscillation power distribution related to the encoding of olfactory information.
[0125] The olfactory function assessment module 350 is used to input the olfactory physiological fingerprint into the trained domain-adaptive olfactory function assessment model to predict the subject's current olfactory sensitivity, discrimination ability, memory retrieval ability, and emotional response intensity.
[0126] Figure 4 The diagram shows a structural schematic of an embodiment of the computer device of the present invention. The specific embodiments of the present invention do not limit the specific implementation of the computer device.
[0127] like Figure 4 As shown, the computer device may include: a processor 402, a communications interface 404, a memory 406, and a communications bus 408.
[0128] The processor 402, communication interface 404, and memory 406 communicate with each other via communication bus 408. Communication interface 404 is used to communicate with other network elements such as clients or other servers. The processor 402 executes program 410, specifically performing the relevant steps in the above-described embodiment of the brain-computer interface-based olfactory fingerprint construction and olfactory function assessment method.
[0129] Specifically, program 410 may include program code that includes computer operation instructions.
[0130] Processor 402 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The computer device includes one or more processors, which may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.
[0131] Memory 406 is used to store program 410. Memory 406 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.
[0132] According to the scheme provided by the present invention, an olfactory stimulus molecular library is constructed based on the individualized information of the test subject; wherein, the individualized information includes the test subject's olfactory receptor genotype data and olfactory preference / aversion memories; the olfactory stimulus molecular library includes pleasantness, intensity, familiarity, complexity characteristics and corresponding olfactory stimulus spectra; olfactory stimuli of preset concentration, flow rate and duration are provided to the test subject's nasal cavity, and background information of the olfactory stimuli is recorded; wherein, the background information includes the semantic label of the odor source, the intensity of emotional induction, the intensity of memory association, or the test subject's association report with the odor source; preset concentrations, flow rates and durations of olfactory stimuli are provided during and after the presentation of olfactory stimuli. Within a given timeframe, brain physiological activity signals of the subject are synchronously acquired via a brain-computer interface cap, which includes a high-density electrode array, an eye-tracking sensor, an electromyography (EMG) sensor, and a near-infrared spectroscopy sensor array. Time-frequency domain analysis is performed on the subject's brain physiological activity signals based on an olfactory stimulus molecular library. An olfactory physiological fingerprint is constructed based on the extracted microstate sequences related to olfactory information encoding, phase-amplitude coupling strength, and high-frequency oscillation power distribution. This olfactory physiological fingerprint is then input into a trained domain-adaptive olfactory function assessment model to predict the subject's current olfactory sensitivity, discrimination ability, memory retrieval ability, and emotional response intensity. This invention acquires the brain's response to olfactory stimuli through a brain-computer interface, reflecting an individual's neural activity patterns under different olfactory stimuli, thereby more accurately assessing olfactory sensitivity, discrimination ability, memory retrieval ability, and emotional response intensity.
[0133] Those skilled in the art will understand that modules in the device of the embodiments can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiments can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components. Except where at least some of such features and / or processes or units are mutually exclusive, any combination of all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or device so disclosed can be employed. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose. Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features from different embodiments are intended to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments can be used in any combination. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims listing several devices, several of these devices may be embodied by the same hardware item. Unless otherwise specified, the steps in the above embodiments should not be construed as limiting the order of execution.
Claims
1. A method for constructing olfactory fingerprints and evaluating olfactory function based on brain-computer interfaces, characterized in that, include: An olfactory stimulus molecular library was constructed based on the individualized information of the test subjects; wherein, the individualized information includes the olfactory receptor genotype data and olfactory preference / dislike memories of the test subjects; the olfactory stimulus molecular library includes pleasantness, intensity, familiarity, complexity features and corresponding olfactory stimulus spectra; An olfactory stimulus of preset concentration, flow rate, and duration is provided into the nasal cavity of the subject, and background information of the olfactory stimulus is recorded; wherein, the background information includes the semantic label of the odor source, the intensity of emotional evokedness, the intensity of memory association, or the subject's association report with the odor source; During the presentation of olfactory stimulation and within a preset time after the stimulation, the brain physiological activity signals of the subject are synchronously collected through the brain-computer interface cap. The brain-computer interface cap includes a high-density electrode array, an eye-tracking sensor, an electromyography sensor, and a near-infrared spectroscopy sensor array. Based on the olfactory stimulation molecule library, the brain physiological activity signals of the test subjects were analyzed in the time and frequency domain. Based on the extracted microstate sequences, phase-amplitude coupling strength and high-frequency oscillation power distribution related to the olfactory information encoding, an olfactory physiological fingerprint was constructed. The olfactory physiological fingerprint is input into a trained domain-adaptive olfactory function assessment model to predict the subject's current olfactory sensitivity, discrimination ability, memory retrieval ability, and emotional response intensity.
2. The method for constructing olfactory fingerprints and evaluating olfactory function based on brain-computer interfaces according to claim 1, characterized in that, When extracting the high-frequency oscillation power distribution related to olfactory information encoding based on the brain physiological activity signals of the test subjects, the high-frequency oscillation power at each time point is calculated; the formula for calculating the high-frequency oscillation power is as follows: in, and It is the preset time range during and after the presentation of olfactory stimuli; and To set the frequency range; It is a probability weighting factor; For frequency The time-frequency distribution function at time point t; Let be the rate of change of power at time t. , For time intervals, For the signal at time point t within the oscillation range [ Total power within; , The number of window functions. For the first Brain physiological activity signals at each sampling point is the window function at time point t.
3. The method for constructing olfactory fingerprints and evaluating olfactory function based on brain-computer interfaces according to claim 1, characterized in that, The method further includes: The dynamic complexity index of a microstate is calculated based on the probability of its occurrence at a given time point and the transition probability of that microstate. The formula for calculating the dynamic complexity index is as follows: in, The number of microstates; It is a sequence of microstates; To start from microstate The probability of transitioning to microstate j; For observation time; For the microstate at time t The probability of occurrence.
4. The method for constructing olfactory fingerprints and evaluating olfactory function based on brain-computer interfaces according to claim 2, characterized in that, The method further includes: The low-frequency phase signal is divided into multiple phase intervals, and the average value of the high-frequency amplitude signal in each phase interval is calculated. Based on the influence of different frequency combinations on the encoding of olfactory information, the coupling index of all possible low-frequency / high-frequency combinations is calculated; The phase / amplitude coupling strength characteristics are improved based on weighting factors and coupling exponents for all possible low / high frequency combinations; where the coupling strength characteristics are: in, To measure the weighting factor of the combination of low frequency i and high frequency j in olfactory perception; Let be the coupling index of the frequency combination (i,j). N is the total number of phase intervals into which the low-frequency phase signal is divided. For a phase signal of a specific low frequency i to fall within the k-th phase interval The average value of the amplitude signal at the corresponding high-frequency j. For a phase signal of a specific low frequency i to fall within the l-th phase interval The average value of the amplitude signal at the corresponding high frequency j; ， ； For falling within the phase interval Number of time points It is the phase signal of low frequency i at time t.
5. The method for constructing olfactory fingerprints and evaluating olfactory function based on brain-computer interfaces according to claim 1, characterized in that, The construction of the olfactory stimulus molecule library based on the individualized information of the test subject further includes: The complete olfactory receptor gene sequence of the subject was obtained using gene sequencing methods. The olfactory receptor gene sequences were compared and analyzed to identify key gene loci and mutation types. The olfactory stimulation molecule library is constructed by associating the gene loci, mutation types, and memory information.
6. The method for constructing olfactory fingerprints and evaluating olfactory function based on brain-computer interfaces according to claim 1, characterized in that, The domain-adaptive olfactory function assessment model includes a physiological fingerprint encoder, a background information encoder, a cross-modal fusion and attention layer, a domain-invariant feature learning layer, and an olfactory function prediction head; The physiological fingerprint encoder includes a micro-state sequence processing module, a phase-amplitude coupling processing module, and a phase-amplitude coupling processing module. The background information encoder includes a text information encoding module and a numerical information encoding module; The cross-modal fusion and attention layer dynamically control the contribution ratio of different modal information through a gated multimodal fusion unit; The domain-invariant feature learning layer includes a feature extractor, a domain discriminator, and a joint optimization objective; The olfactory function prediction head includes a multi-task predictor composed of multiple independent or shared underlying fully connected neural network branches, with each neural network branch used to predict an olfactory function index.
7. The method for constructing olfactory fingerprints and evaluating olfactory function based on brain-computer interfaces according to claim 6, characterized in that, The microstate sequence processing module adopts a Bi-GRU bidirectional gated cyclic unit network layer, and the GRU unit in the Bi-GRU bidirectional gated cyclic unit network layer includes an update gate and a reset gate. The phase-amplitude coupling processing module includes a one-dimensional convolutional neural network layer and a two-dimensional convolutional neural network layer; The gated multimodal fusion unit adopts a dynamic routing attention mechanism. The dynamic routing mechanism allocates the initial fusion features to different capsules by iteratively updating the routing weights. Each capsule represents an olfactory feature mode. The outputs of all capsules are concatenated to obtain cross-modal fusion features. The neighborhood discriminator uses a GRL gradient inversion layer to connect the feature extractor and the neighborhood discriminator; wherein, the GRL gradient inversion layer keeps the input unchanged during forward propagation and inverts the gradient during backward propagation; The olfactory function prediction head uses different activation functions for different olfactory function indicators; a linear activation function is used for olfactory sensitivity and discrimination ability, and a softmax activation function is used for emotional response intensity level.
8. The method for constructing olfactory fingerprints and evaluating olfactory function based on brain-computer interface according to claim 4, characterized in that, The step of performing time-frequency domain analysis of the subject's brain physiological activity signals based on the olfactory stimulation molecular library further includes: Based on the characteristics of pleasure, intensity, familiarity, and complexity of the olfactory stimulus molecular library and the olfactory stimulus spectrum, the olfactory stimuli were grouped, and the brain physiological activity signals corresponding to each group of stimuli were decomposed in time and frequency. In each olfactory stimulation group, the number of clusters or convergence criteria are guided by the complexity characteristics of each stimulus in the olfactory stimulation molecule library to obtain a prototype map of brain microstates under that stimulation group; the brain topography map at each time point is matched with the obtained prototype map of microstates, and the duration, frequency of occurrence and transition probability between microstates under different stimulation groups are calculated to obtain a microstate sequence. Before calculating the average value of the high-frequency amplitude signal in each phase interval, the low-frequency phase signal and the high-frequency amplitude signal are extracted by Hilbert transform. Before calculating the high-frequency oscillation power at each time point, the power values of the high-gamma band under different olfactory stimulation groups are calculated and the distribution of the high-density electrode arrays in different brain-computer interface caps is analyzed to obtain the high-frequency oscillation power distribution related to the emotional induction intensity and memory association intensity of each stimulus in the olfactory stimulation molecular library.
9. A brain-computer interface-based olfactory fingerprint construction and olfactory function assessment device, characterized in that, include: The olfactory stimulus molecular library construction module is used to construct an olfactory stimulus molecular library based on the individualized information of the test subjects; wherein, the individualized information includes the olfactory receptor genotype data and olfactory preference / aversion memory of the test subjects; the olfactory stimulus molecular library includes pleasantness, intensity, familiarity, complexity features and corresponding olfactory stimulus spectra; The olfactory stimulus presentation and background information recording module is used to provide the subject with an olfactory stimulus source of preset concentration, flow rate and duration in the nasal cavity, and record the background information of the olfactory stimulus source; wherein, the background information includes the semantic label of the odor source, the intensity of emotional induction, the intensity of memory association or the subject's association report with the odor source; The brain physiological activity signal acquisition module is used to synchronously acquire the brain physiological activity signals of the subject through a brain-computer interface cap during the presentation of olfactory stimulation and within a preset time after the presentation of stimulation. The brain-computer interface cap includes a high-density electrode array, an eye-tracking sensor, an electromyography sensor, and a near-infrared spectroscopy sensor array. The olfactory physiological fingerprint construction module is used to perform time-frequency domain analysis on the brain physiological activity signals of the subject based on the olfactory stimulation molecule library, and to construct the olfactory physiological fingerprint based on the extracted microstate sequence, phase-amplitude coupling strength and high-frequency oscillation power distribution related to the encoding of olfactory information. The olfactory function assessment module is used to input the olfactory physiological fingerprint into a trained domain-adaptive olfactory function assessment model to predict the subject's current olfactory sensitivity, discrimination ability, memory retrieval ability, and emotional response intensity.
10. A computer device, comprising: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction, which causes the processor to perform the operations corresponding to the above-described brain-computer interface-based olfactory fingerprint construction and olfactory function assessment method.