Wearable olfactory stimulation modulation device and method

Abstract

This invention provides a wearable olfactory stimulation modulation device and method, belonging to the field of wearable intelligent modulation technology. The method includes preprocessing the original signal, removing interference artifacts and environmental noise, extracting effective EEG features, and using a lightweight depression recognition model for inference and calculation to determine the severity of the user's current depressive disorder. Based on the severity of depression, an adaptive adjustment algorithm generates olfactory stimulation control commands, specifying parameters such as odor type and delivery intensity. This invention, employing the aforementioned wearable olfactory stimulation modulation device and method, solves the limitations of existing clinical modulation methods for depressive disorders and the drawbacks of existing olfactory modulation-related devices.

Description

Technical Field

[0001] This invention relates to the field of wearable intelligent control technology, and in particular to a wearable olfactory stimulation control device and method. Background Technology

[0002] Current clinical and daily management methods for depressive disorders have several limitations. Drug therapy exhibits individual differences in response, with some patients failing to achieve clear therapeutic effects, and is accompanied by various physical side effects; long-term use also carries the risk of medication dependence. Psychological intervention therapy is highly dependent on the time and expertise of professional medical staff, has high treatment costs, and some patients have low acceptance of psychological intervention, making it difficult to complete a full treatment cycle. Traditional physical therapy methods have high operational barriers, involve certain physical trauma and safety risks during the treatment process, and are not suitable for patients' daily home use needs.

[0003] Existing olfactory modulation devices for depressive disorders have significant technical shortcomings. Most devices are non-wearable, fixed devices, bulky and poorly portable, failing to meet the needs of patients in daily life, work, and other scenarios. Some devices only have EEG data acquisition capabilities, unable to simultaneously achieve olfactory stimulation modulation, thus failing to complete the entire detection and treatment process. Furthermore, the depression recognition algorithms of some devices require uploading the collected EEG data to the cloud for processing; data transmission is prone to frame drops and delays, preventing real-time recognition. Additionally, existing devices pose a risk of user physiological data leakage, compromising the accuracy and real-time performance of the algorithm. Moreover, the olfactory stimulation modules of existing devices cannot achieve precise control of essential oil delivery; the diffusion loss during odor delivery is significant, preventing targeted olfactory stimulation and the ability to adjust stimulation parameters based on the user's real-time state, thus failing to form a complete regulatory loop and making it difficult to guarantee the effectiveness of depression control.

[0004] Therefore, there is an urgent need for a wearable olfactory stimulation modulation device and method to solve the above problems. Summary of the Invention

[0005] The purpose of this invention is to provide a wearable olfactory stimulation modulation device and method, which solves the limitations of existing clinical modulation methods for depressive disorders, and the problems of poor portability, incomplete functions, data transmission and security issues, and inaccurate olfactory stimulation control of existing olfactory modulation related devices.

[0006] To achieve the above objectives, the present invention provides a wearable olfactory stimulation modulation method, comprising the following steps: S1. The raw EEG signal is preprocessed to remove interference artifacts and environmental noise, and effective EEG features are extracted. A lightweight depression recognition model based on the feature map pruning method COPruner is used to infer and calculate the effective EEG features to determine the severity of the user's current depressive disorder. S2. Based on the severity of depression, an adaptive adjustment algorithm is used to generate olfactory stimulation control instructions, using EEG depression biomarkers as the regulation target to determine the odor type, delivery intensity, stimulation duration and stimulation frequency.

[0007] Preferably, the preprocessing operations of the raw EEG signal in S1 include baseline correction, power frequency filtering, removal of electrooculography artifacts and removal of electromyography artifacts. The preprocessed EEG signal is then subjected to feature extraction to obtain EEG features related to the severity of depression. The EEG feature indicators include the frequency domain features, time domain features and nonlinear features of the EEG signal.

[0008] Preferably, the method for constructing the mild depression identification model in S1 includes: S11. Channel Search Space Constraint: For the ResNet 3-layer basic model, the number of channels retained in each convolutional layer is limited to a preset discrete set. Within this framework, the combinatorial search space is narrowed down through constraints; among which, For the original convolutional layer j Number of channels in the layer Reserve the maximum percentage for the preset channels; S12, Pruning Structure Optimization Search: Initialize the pruning structure set, pre-train the base model weights, take maximizing classification accuracy as the optimization goal, combine the feature map pruning method COPruner based on group collaboration optimization, iteratively update the pruning structure set, and select the pruning structure with the best fitness. S13. Model fine-tuning and deployment: Fine-tune the lightweight model corresponding to the optimal pruning structure, solidify the model weights, and deploy it to the embedded microcontroller for real-time depression detection and severity determination.

[0009] The preferred quantitative formula for the optimization objective is: ; in, This is the network structure of the pruned model. This is the pruned CNN model. For the pruned model on the training set Weights for training or fine-tuning For the model on the test set The accuracy rate of depression detection on the device; Introducing channel constraints, the optimization objective is constrained as follows: ; in, For the model after pruning i Number of feature map channels in the layer L This represents the total number of convolutional layers in the CNN model.

[0010] Preferably, in the feature map pruning method COPruner based on group collaborative optimization, the pruning structure set is iteratively updated through working agents, selecting agents, and re-initializing agents to select the pruning structure with the best fitness. The working agent is used to generate new structure candidates for each pruning structure. The quantization generation formula for the new structure candidates is as follows: ; in, For the generated new structure candidate i Number of channels in the layer The original pruning structure i Number of channels in the layer r A random number in the range of 0-1. The current optimal pruning structure is the i Number of channels in the layer A rounding function that returns the value within a preset set of discrete channels that is closest to the input value; The working agent determines whether to replace the original structure with the generated structure candidate based on fitness. The formula for quantifying fitness is as follows: ; in, For the first j A pruning structure Adaptability, Indicates the first A pruned network Indicates the first A pruning network The corresponding weights This represents the test set. Represents the training set; The agent selection method is used to select pruning structures based on fitness-related probabilities, generating better pruning structure candidates. The quantification formula for the selection probability is as follows: ; in, For the first j The selection probability of a pruned structure This represents the maximum fitness value in the current pruning structure set; The reinitialization agent is used to reinitialize pruned structures that have not been optimized after exceeding a preset number of updates, thus preventing the search from getting trapped in local optima.

[0011] A wearable olfactory stimulation modulation device includes a wearable helmet body, an electroencephalogram (EEG) signal acquisition unit, a central processing and control unit, an odor generation unit, and an odor-directed delivery unit. A head-fixing component is located inside the wearable helmet body, and several ventilation holes are provided on the outer surface of the helmet body. A control mounting position is located on the top of the helmet body, and odor bottle mounting positions are symmetrically arranged on both sides of the helmet body. The EEG signal acquisition unit includes multiple sets of EEG electrodes embedded inside the wearable helmet body, which maintain stable contact with the user's scalp through the head-fixing component. The central processing and control unit is installed... In the control mounting position, the central processing and control unit has a built-in embedded microcontroller; the odor generating unit includes a detachable olfaction bottle and a corresponding micro metering pump. The detachable olfaction bottle includes a working olfaction bottle group and a spare olfaction bottle group. The detachable olfaction bottle is installed in the olfaction bottle mounting position. The air inlet of the micro metering pump is sealed and connected to the air outlet of the corresponding detachable olfaction bottle. The controlled end of the micro metering pump is electrically connected to the control output end of the central processing and control unit; the air inlet of the odor directional delivery unit is connected to the air outlet of the micro metering pump. The air outlet of the odor directional delivery unit is equipped with a nozzle whose spatial position can be flexibly adjusted. The nozzle is positioned towards the user's nasal cavity area.

[0012] Preferably, the head fixation component includes an annular fixation airbag adapted to the user's head circumference and silicone pads fixedly installed on the inner annular surface of the annular fixation airbag. The annular fixation airbag is arranged around the inner edge of the wearable helmet body, and the silicone pads are evenly distributed on the inner annular surface of the annular fixation airbag.

[0013] Preferably, the EEG electrodes are placed on both sides of the head, the forehead, the top of the head, and the occipital region of the wearable helmet body, and the rear end of each EEG electrode is provided with an elastic buffer.

[0014] Preferably, the detachable sniff bottle has a sealing membrane at the air outlet, and the air inlet of the micro metering pump is provided with a puncture protrusion and a seal. The detachable sniff bottle contains an essential oil medium, which includes valence-inducing and arousal-inducing scents.

[0015] Preferably, the odor delivery unit includes a metal-shaped flexible tube, one end of which is sealed and connected to the outlet of a micro metering pump, and the other end of which is fixedly connected to a nozzle.

[0016] Therefore, the present invention employs the above-described wearable olfactory stimulation modulation device and method, and the technical effects are as follows: 1. By integrating a helmet-style design with an adaptive head fixation structure, the problems of existing depression control devices being large, inconvenient to carry, and unsuitable for daily wear and use have been solved.

[0017] 2. By using local real-time processing of EEG signals and a mild depression recognition model, the problems of cloud computing latency, easy data leakage, and untimely recognition of depression state are solved.

[0018] 3. By using a micro metering pump and a directional atomization delivery structure, the problems of uncontrollable essential oil delivery, large loss of aroma diffusion, and poor targeted stimulation effect are solved.

[0019] 4. By using real-time EEG feedback and adaptive olfactory stimulation closed-loop regulation, the problems of fixed stimulation parameters, lack of dynamic adjustment, and unstable depression regulation effect of traditional devices are solved. Attached Figure Description

[0020] Figure 1 This is a schematic diagram showing the orientation of a wearable olfactory stimulation modulation device according to the present invention. Figure 2 This is a schematic diagram of the overall orientation of a wearable olfactory stimulation modulation device according to the present invention. Figure 3 This is a schematic diagram of the orientation of a wearable olfactory stimulation modulation device according to the present invention. Figure 4 This is a schematic diagram of the orientation of a wearable olfactory stimulation modulation device according to the present invention. Figure 5 This is a schematic diagram of the orientation of a wearable olfactory stimulation modulation device according to the present invention. Figure 6 This is a schematic diagram of the feature map pruning method COPruner based on group collaboration optimization in an embodiment of the present invention; Figure 7 This is a schematic diagram of the variable olfactory stimulation paradigm in an embodiment of the present invention.

[0021] Figure Labels 1. Wearable helmet body; 101. Ventilation vent; 102. Control mounting position; 103. Smell bottle mounting position; 11. Head fixation assembly; 111. Circular fixation airbag; 112. Silicone pad; 2. EEG electrodes; 3. Detachable smell bottle; 301. Miniature metering pump; 302. Nozzle; 4. Metal shaping hose. Detailed Implementation

[0022] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0023] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0024] Example 1 like Figures 1-5As shown, this invention provides a wearable olfactory stimulation modulation device, including a wearable helmet body 1, an electroencephalogram (EEG) signal acquisition unit, a central processing and control unit, an odor generation unit, and an odor-directed delivery unit. The wearable helmet body is integrally molded from lightweight materials, integrating the EEG signal acquisition unit, central processing and control unit, odor generation unit, and odor-directed delivery unit. It can operate independently without external devices, making it suitable for users' daily scenarios such as home, travel, and work. The helmet body surface is evenly distributed with ventilation holes to improve air circulation and heat dissipation, avoiding stuffiness and discomfort. The head fixation component adopts a ring-shaped fixation airbag combined with silicone pads, which can adaptively adjust the fit to accommodate different head circumferences, ensuring wearing stability and contact comfort, and optimizing the long-term wearing experience.

[0025] The wearable helmet body 1 serves as the support base for each functional unit. A head-fixing component 11, which can adaptively adjust its fit to the user's head, is located on the inner side of the wearable helmet body 1. Several ventilation holes 101 are provided on the outer surface of the wearable helmet body 1. A control mounting position 102 is located on the top of the wearable helmet body 1, and snorkel mounting positions 103 are symmetrically arranged on both sides of the wearable helmet body 1. The head-fixing component 11 includes an annular fixing airbag 111 adapted to the user's head circumference and several silicone pads 112 fixedly mounted on the inner annular surface of the annular fixing airbag 111. The annular fixing airbag 111 is arranged around the inner edge of the wearable helmet body 1. The annular fixing airbag 111 can adjust its expansion by inflating and deflating, achieving adaptive adjustment of the fit between the wearable helmet body 1 and the user's head. The silicone pads 112 are evenly distributed on the inner annular surface of the annular fixing airbag 111 to improve wearing comfort. Simultaneously, the expansion force of the annular fixing airbag 111 helps the EEG electrodes 2 to fit tightly against the user's scalp.

[0026] The annular fixing airbag 111 adjusts its expansion by inflating and deflating to fit different users' head circumferences, achieving adaptive adjustment of the fit between the wearable helmet body 1 and the user's head, ensuring the fit and stability of the device. The silicone pads 112 are evenly distributed on the inner ring surface of the annular fixing airbag 111, improving the user's contact comfort during wear. At the same time, the expansion force of the annular fixing airbag 111 helps the EEG electrodes 2 to fit tightly against the user's scalp, further improving the stability and accuracy of EEG signal acquisition.

[0027] The EEG signal acquisition unit includes multiple sets of EEG electrodes 2 embedded inside the wearable helmet body 1. The signal output terminals of the EEG electrodes 2 are electrically connected to the signal input terminals of the central processing and control unit. The EEG electrodes 2 are used to acquire the user's EEG signals in real time and transmit them to the central processing and control unit. The head fixation component 11 is used to help the EEG electrodes 2 maintain stable contact with the user's scalp. The EEG electrodes 2 adopt a four-channel layout. The multiple sets of EEG electrodes 2 are respectively arranged in accordance with the international 10-20 EEG electrode arrangement standard in the two sides of the head, the forehead area, the top of the head area, and the occipital area of ​​the wearable helmet body 1. Each set of EEG electrodes 2 has an elastic buffer at its rear end. The elastic buffer is used to provide a pre-tightening force towards the user's scalp for the EEG electrodes 2, improving the stability and accuracy of EEG signal acquisition.

[0028] The central processing and control unit is fixedly installed on the control mounting position 102 on the top of the wearable helmet body 1. The central processing and control unit has an embedded microcontroller built in, which is preloaded with a depression state recognition algorithm and an adaptive adjustment algorithm. The central processing and control unit is used to preprocess and extract features from the received EEG signals, determine the severity of the user's depression disorder through the depression state recognition algorithm, and generate corresponding olfactory stimulation control commands based on the determination results through the adaptive adjustment algorithm.

[0029] The olfactory generation unit includes two sets of detachable sniff bottles 3 and miniature metering pumps 301 corresponding to each detachable sniff bottle 3. One set of detachable sniff bottles 3 is the working sniff bottle set, and the other set is the spare sniff bottle set. The detachable sniff bottles 3 are detachably installed on the sniff bottle mounting positions 103 on both sides of the wearable helmet body 1. The detachable sniff bottles 3 are used to store mood-regulating essential oil media. The air inlet of the miniature metering pump 301 is sealed and connected to the air outlet of the corresponding detachable sniff bottle 3. The controlled end of the miniature metering pump 301 is electrically connected to the control output end of the central processing and control unit. The miniature metering pump 301 is used to complete the precise metering delivery of essential oil media according to olfactory stimulation control commands. The air inlet of the odor-directed delivery unit is connected to the air outlet of the miniature metering pump 301. The air outlet of the odor-directed delivery unit is equipped with a nozzle 302 whose spatial position can be flexibly adjusted. The nozzle 302 is arranged facing the user's nasal cavity area to directionally deliver the aroma emitted by the metered essential oil media to the user's nasal cavity. The detachable sniff bottle 3 has a sealing membrane at its air outlet. The inlet of the micro metering pump 301 is equipped with a puncture protrusion and a seal. The puncture protrusion is used to puncture the sealing membrane when the detachable sniff bottle 3 is installed in the sniff bottle mounting position 103. The seal is used to wrap the outer wall of the air outlet of the detachable sniff bottle 3, so as to achieve the leakage-free delivery of the odor emitted by the essential oil medium.

[0030] The sealing membrane at the air outlet of the detachable sniff bottle 3 can prevent leakage and evaporation of the internal essential oil medium when the detachable sniff bottle 3 is not installed, thus extending the storage period of the essential oil medium. The puncture protrusion at the air inlet of the micro metering pump 301 can puncture the sealing membrane when the detachable sniff bottle 3 is installed to the sniff bottle installation position 103, thereby enabling the conduction of the essential oil medium and simplifying the installation operation of the detachable sniff bottle 3. The seal can wrap around the outer wall of the air outlet of the detachable sniff bottle 3 to prevent leakage during the transportation of the essential oil medium and ensure the sealing and safety of the essential oil transportation process.

[0031] The odor delivery unit includes a metal shaped hose 4. One end of the metal shaped hose 4 is sealed and connected to the outlet end of the micro metering pump 301. The other end of the metal shaped hose 4 is fixedly connected to the nozzle 302. The metal shaped hose 4 can be bent at will and maintain its shaped state, thereby driving the nozzle 302 to be adjusted to the target position aimed at the user's nasal cavity.

[0032] The metal-shaped flexible tube 4 is connected at both ends to the outlet of the micro metering pump 301 and the nozzle 302, respectively, enabling a stable delivery of the essential oil medium from the micro metering pump 301 to the nozzle 302. The metal-shaped flexible tube 4 can be bent arbitrarily and maintain its shape, allowing the nozzle 302 to flexibly adjust its spatial position to suit different users' facial features. This ensures that the nozzle 302 can be precisely aimed at the user's nasal cavity area, guaranteeing the accuracy of the directional delivery of the essential oil medium. The nozzle 302 is an atomizing nozzle used to atomize the essential oil medium and release it directionally into the user's nasal cavity, improving the uniformity of olfactory stimulation and absorption efficiency.

[0033] The EEG signal acquisition unit is also used to continuously acquire the user's real-time EEG signals during the implementation of olfactory stimulation, so that the central processing and control unit can dynamically adjust the olfactory stimulation control instructions according to the real-time updated EEG signals, forming a closed-loop regulation based on the severity of depressive disorders based on real-time feedback of EEG signals.

[0034] The wearable helmet body 1 integrates all functional units, achieving an integrated wearable design that allows for independent operation without external devices, adapting to various daily usage scenarios and enhancing portability. The head-fixing component 11 is adjustable to fit the user's head, ensuring stability while also maintaining stable contact between the EEG electrodes 2 and the user's scalp, improving the reliability of EEG signal acquisition. Ventilation holes 101 enhance breathability and heat dissipation during helmet wear, optimizing long-term comfort. The EEG electrodes 2 collect the user's EEG signals in real time, providing a data foundation for assessing depressive states. The central processing and control unit performs EEG signal processing and analysis, determining the severity of depression. The system integrates severity assessment and control command generation, achieving integrated processing from detection to regulation. A detachable olfactory bottle 3, paired with a micro-quantitative pump 301, enables the storage and precise delivery of essential oil media. The setup of two sets of olfactory bottles—one for operation and one for backup—ensures continuous device operation and prevents interruption of regulation due to essential oil depletion. The nozzle 302 is positioned towards the nasal cavity, directing the essential oil media to the user's nasal cavity, reducing diffusion loss during delivery and enhancing the effectiveness of olfactory stimulation. Continuous acquisition of real-time EEG signals and dynamic adjustment of control commands form a closed-loop regulation system based on EEG signal feedback. This allows for real-time adjustment of the regulation strategy according to changes in the user's state, ensuring the continuous effectiveness of depression regulation.

[0035] The wearable helmet body 1 is made of lightweight ABS engineering plastic through a single molding process, which reduces the overall weight of the device and improves its ease of wear. At the same time, the single-molded structure ensures the structural strength and stability of the wearable helmet body 1. Ventilation holes 101 are evenly distributed along the surface of the wearable helmet body 1, which can improve air circulation during wear, enhance heat dissipation, avoid stuffiness and discomfort during prolonged wear, and optimize the user's wearing experience.

[0036] The working olfactory bottle set stores essential oil media for daily regulation, while the spare olfactory bottle set stores essential oil media for emergency regulation or replacement fragrance essential oil media. The central processing and control unit can switch the working status of the micro metering pumps 301 corresponding to the two sets of detachable olfactory bottles 3 according to regulation needs. The essential oil media are divided into two categories: valence-based emotional induction scents and arousal-based emotional induction scents. The valence-based emotional induction scents are pleasant scents such as bergamot and lavender, while the arousal-based emotional induction scents are arousing scents such as lemon and rosemary. The duration of a single scent stimulation is preset to 60 seconds. It can target the characteristics of low arousal and low pleasure in depressive disorders, simultaneously covering the regulation needs of both the valence and arousal dimensions, improving the dimensions of depression regulation, and enhancing the comprehensiveness and effectiveness of regulation.

[0037] The inner side of the wearable helmet body 1 is also equipped with a rechargeable power module, which is electrically connected to the EEG signal acquisition unit, the central processing and control unit, and the micro quantitative pump 301 to provide power for the entire device.

[0038] The central processing and control unit incorporates a multivariate olfactory stimulation paradigm execution logic. This paradigm first acquires the user's subjective evaluation of the emotional dimensions evoked by the odor, and then executes odor-valence dimension experiments and odor-arousal dimension experiments based on the evaluation results. In the odor-valence dimension experiment, a pleasant odor with equal probability is selected as the bias stimulus, and in the odor-arousal dimension experiment, a relaxing odor with equal probability is selected as the bias stimulus. Both types of experiments use distilled water or air as the standard stimulus. The setting of bias and standard stimuli, combined with the user's key press response, can simultaneously acquire the user's subjective behavioral data and objective EEG data, enriching the data dimensions for determining depressive states, while ensuring the user's attention is focused during stimulation, thus enhancing the effectiveness of olfactory stimulation.

[0039] Olfactory stimulation modulation is achieved through the aforementioned wearable olfactory stimulation modulation device, including: S1. Real-time acquisition of EEG signals: The user's raw EEG signals are acquired in real time through the EEG signal acquisition unit on the wearable helmet body 1, and the raw EEG signals are transmitted to the central processing and control unit. S2. EEG Signal Processing and Severity Assessment of Depressive Disorder: The central processing and control unit preprocesses the received raw EEG signals to remove interference artifacts and environmental noise, extracts effective EEG features, and performs inference calculations on the effective EEG features using a lightweight depression detection model built by the pre-loaded group collaboration optimization-based feature map pruning method COPruner to determine the user's current depression severity. Among the effective EEG features are olfactory evoked EEG features, which are extracted by the olfactory EEG feature extraction module. Based on the extracted olfactory evoked EEG features, the user's degree and state of depressive disorder are assessed using a group collaboration optimization-based depression detection method. Preprocessing and feature extraction of raw EEG signals can remove interference components from the signals and extract effective EEG features, ensuring the accuracy of the determination of the severity of depression. Among them, the extraction of olfactory-evoked EEG features can further strengthen the correlation between EEG features and olfactory stimulation and improve the targeting of depression detection. S3. Adaptive generation of olfactory stimulation control commands: The central processing and control unit generates corresponding olfactory stimulation control commands based on the severity of the depressive disorder, using a pre-loaded adaptive adjustment algorithm. Based on the mapping relationship between EEG biomarkers and olfactory stimulation parameters (Mb=f(Q,M,T,F,t)), it determines the appropriate essential oil medium type, quantitative delivery parameters, and working sequence, while simultaneously controlling the switching between the working and standby olfactory bottle groups. In the mapping relationship, Mb is the depressive biomarker of the EEG signal, Q is the odor type, M is the odor intensity, T is the stimulation duration, F is the stimulation frequency, and t is the time variable. The functional expression of the mapping relationship is determined through multivariate regression analysis, and a PID algorithm is used to optimize the four core stimulation parameters: odor type Q, odor intensity M, stimulation duration T, and stimulation frequency F. Based on a multivariate olfactory stimulation paradigm, and considering the characteristics of low arousal and low pleasure in depressive disorders, it balances valence and arousal dimensions for dual-dimensional regulation, selecting appropriate valence-based or arousal-based emotion-inducing odors. The duration of a single odor stimulation segment is set to 60 seconds. Based on the severity of depression, corresponding olfactory stimulation control instructions are generated, which can match the olfactory stimulation parameters with the user's current depressive state to achieve adaptive matching of the regulation strategy. The stimulation parameters are determined based on the mapping relationship between EEG biomarkers and olfactory stimulation parameters, which can achieve quantitative control of the regulation process. At the same time, it takes into account the dual-dimensional regulation of valence and arousal, which can adapt to the emotional characteristics of depressive disorders and improve the comprehensiveness of regulation. S4. Olfactory Stimulation Execution and Targeted Delivery: The central processing and control unit drives the corresponding micro metering pump 301 to operate according to the olfactory stimulation control command, and controls the essential oil medium in the corresponding detachable olfactory bottle 3 to complete the metered delivery according to the set parameters. The delivered essential oil medium is directed to the user's nasal cavity through the nozzle 302 of the odor targeted delivery unit to implement targeted olfactory stimulation for the user. Based on the multivariate olfactory stimulation paradigm, when performing odor-valence dimension experiment or odor-arousal dimension experiment, the biased stimulus and standard stimulus are presented to the user in a multivariate random manner, and the user's key press response data and real-time EEG signals are collected simultaneously. Driven by control commands, the micro metering pump 301 is operated to complete the quantitative delivery and directional release of essential oil media, thereby achieving targeted olfactory stimulation of users and ensuring the precise implementation of the regulatory effect. The presentation of variable random stimulation based on the multivariate olfactory stimulation paradigm can reduce interference between dimensions and improve the effectiveness of experimental data. S5. Closed-loop feedback control: Throughout the entire process of olfactory stimulation, the EEG signal acquisition unit continuously acquires the user's real-time EEG signals and repeats steps S2 to S4. Based on the dynamic changes in the severity of the user's depressive disorder, the olfactory stimulation control commands are adjusted in real time to achieve closed-loop adaptive control of the depressive state throughout the entire process. Through the linkage of the olfactory stimulation module, the olfactory EEG feature extraction module, the depressive disorder state detection module, and the olfactory stimulation intelligent control module, the olfactory stimulation parameters are continuously optimized to complete the closed-loop operation of acquisition-identification-control-stimulation-reacquisition. By continuously collecting real-time EEG signals and repeatedly executing corresponding steps, a closed-loop adaptive regulation of the entire process of depressive state can be achieved. The regulation strategy can be adjusted in real time according to the dynamic changes of the user's depressive state to ensure the continuous effectiveness of depression regulation. The coordinated operation of multiple modules can improve the closed-loop regulation system and continuously optimize the regulation effect.

[0040] Preprocessing of the raw EEG signals includes baseline correction, power frequency filtering, removal of electrooculography (EOG) artifacts, and removal of electromyography (EMG) artifacts. The preprocessed EEG signals are then used for feature extraction to obtain EEG characteristic indicators related to the severity of depression. These indicators include frequency domain features, time domain features, and nonlinear features of the EEG signals. Preprocessing operations such as baseline correction, power frequency filtering, EOG artifact removal, and EMG artifact removal effectively remove environmental interference and physiological artifacts from the raw EEG signals, improving the signal-to-noise ratio. Extracting frequency domain features, time domain features, and nonlinear features from the preprocessed EEG signals comprehensively reflects changes in the user's EEG signals, providing multi-dimensional feature evidence for determining the severity of depression and further improving the accuracy of depression severity assessment.

[0041] The central processing and control unit stores a mapping table between the severity of depression and olfactory stimulation parameters. This table can match corresponding essential oil media combinations and quantitative delivery parameters for different levels of depression severity, enabling rapid matching of control parameters and improving the device's response speed. The mapping table can be optimized and updated based on the user's historical control data and real-time control effects, allowing the control strategy to be adapted to individual user differences, achieving personalized olfactory stimulation control, and improving the adaptability and effectiveness of depression control.

[0042] The central processing and control unit judges the changing trend of the user's depressive state based on continuously collected real-time EEG signals and can adapt the corresponding regulation strategy in advance. When the user's depression severity drops below the preset safety threshold, the essential oil delivery volume and operating frequency of the micro metering pump 301 are gradually reduced to avoid overstimulation and adapt to the user's gradually improving state. When the user's depression severity increases, the essential oil delivery volume and operating frequency of the micro metering pump 301 are increased accordingly to enhance the regulation effect in a timely manner, realize dynamic adaptive closed-loop regulation, and ensure the adaptability and effectiveness of the regulation process.

[0043] like Figure 6 As shown, the algorithm for identifying the severity of depressive disorders employs a mild depression identification model constructed using the COPruner feature map pruning method based on collaborative group optimization. The construction method for the mild depression identification model includes: S21. Channel Search Space Constraint: For the ResNet 3-layer basic model, the number of channels retained in each convolutional layer is limited to a preset discrete set. Within this framework, the combinatorial search space is narrowed down through constraints; among which, For the original convolutional layer j Number of channels in the layer Reserve the maximum percentage for the preset channels; S22. Pruning Structure Optimization Search: Initialize the pruning structure set, pre-train the base model weights, take maximizing classification accuracy as the optimization goal, combine the feature map pruning method COPruner based on group collaboration optimization, iteratively update the pruning structure set, and select the pruning structure with the best fitness. S23. Model fine-tuning and deployment: Fine-tune the lightweight model corresponding to the optimal pruning structure, solidify the model weights, and deploy it to the embedded microcontroller for real-time depression detection and severity determination.

[0044] First, the combinatorial search space is reduced by limiting the number of channels retained by the convolutional layers to a preset discrete space. The number of channels retained by each layer is limited to a certain value. ,in, For the original convolutional layer j Number of channels in the layer Set a preset upper limit for the channel retention ratio, with a value ranging from 10% to 100%, and This is shared across all convolutional layers; the search for the optimal pruning structure is then formalized into an optimization problem that maximizes classification accuracy, with the optimization objective quantified as follows: ; in, This is the network structure of the pruned model. This is the pruned CNN model. For the pruned model on the training set Weights for training / fine-tuning For the model on the test set The accuracy rate of depression detection on the device; Introducing channel constraints, the optimization objective is constrained as follows: ; in, For the model after pruning i Number of feature map channels in the layer Ldenoted as the total number of convolutional layers in the CNN model. The constrained optimization problem is solved using the COPruner feature map pruning method, which is based on collaborative group optimization. The pruning structure set is iteratively updated through working agents, agent selection, and agent re-initialization to select the pruning structure with optimal fitness. Finally, the model corresponding to the optimal pruning structure is fine-tuned to obtain a lightweight ResNet 3-layer depression recognition model deployed on an embedded microcontroller.

[0045] The adaptive regulation employs a proportional-integral-derivative (PID) control algorithm. The device constructs a closed-loop control logic based on a multivariate olfactory stimulation paradigm, comprising an olfactory stimulation module, an olfactory EEG feature extraction module, a depressive disorder state detection module, and an intelligent olfactory stimulation control module. The olfactory stimulation module inputs odors to the user and acquires EEG signals. The olfactory EEG feature extraction module extracts olfactory-evoked EEG features from the acquired cortical EEG signals. The depressive disorder state detection module assesses the degree and state of depressive disorder using olfactory-evoked EEG features and a group-based collaborative optimization-based depression detection method. The intelligent olfactory stimulation control module, based on the detected degree of depression, adopts a PID control strategy to adaptively output olfactory stimulation parameters, achieving closed-loop control of the depressive disorder state. The PID control algorithm uses the depression-related EEG signals... Using the biomarker Mb as the regulatory target, the output control command is optimized based on the mapping relationship between the EEG biomarker and olfactory stimulation parameters, Mb=f(Q,M,T,F,t), where Q is the odor type, M is the odor intensity, T is the stimulation duration, F is the stimulation frequency, and t is the time variable. This is used to adjust the operating speed, working duration, and start / stop frequency of the micro metering pump 301 according to the identified severity of depression, thereby achieving precise quantitative control of the essential oil medium delivery volume, release duration, and release frequency. The mapping relationship is fitted and solved through multiple regression analysis. By changing the odor type Q, odor intensity M, stimulation duration T, and stimulation frequency F and recording the corresponding EEG biomarker Mb, the dominant or recessive expression of the function f() is determined. The four stimulation parameters are continuously optimized through a PID algorithm to achieve quantitative regulation of depressive disorders.

[0046] The lightweight depression detection model constructed by the feature map pruning method based on group collaboration optimization can complete real-time inference operations locally on an embedded microcontroller, without uploading EEG data to the cloud for processing, avoiding frame dropping and latency problems during data transmission. At the same time, it eliminates the risk of leakage of users' physiological data during the transmission process, improving the real-time and accuracy of the determination of the severity of depression; by iteratively updating the pruning structure set through the collaborative optimization algorithm, the pruning structure with the optimal fitness can be selected, ensuring the detection accuracy of the model while compressing the model volume to adapt to the deployment requirements of embedded devices; the multi-module linkage control logic based on the variable olfactory stimulation paradigm can achieve the full-process linkage of olfactory stimulation, EEG feature extraction, depression state detection and intelligent control, improving the closed-loop control system of the device; the PID control algorithm takes the depression biomarker of the EEG signal as the control target, optimizes the output control instruction based on the mapping relationship between the EEG biomarker and the olfactory stimulation parameters, and correspondingly adjusts the operating state of the micro quantitative pump 301 to achieve precise quantitative control of the delivery volume, release duration and release frequency of the essential oil medium, so that the olfactory stimulation parameters can match the user's depression state and improve the adaptability of the control; by fitting the mapping relationship through multiple regression analysis, the association logic between the olfactory stimulation parameters and the EEG biomarker can be clarified, and the stimulation parameters can be continuously optimized in combination with the PID algorithm to achieve quantitative control of depressive disorders and improve the stability of the control effect.

[0047] In the COPruner of the feature map pruning method based on group collaboration optimization, the working agent is used to generate new structure candidates for each pruning structure. The quantization generation formula for the new structure candidates is: ; Where is the number of channels in the i th layer of the generated new structure candidate, is the number of channels in the i th layer of the original pruning structure, r is a random number within the range of 0 - 1, is the number of channels in the i th layer of the current optimal pruning structure, is the rounding function that returns the value within the preset discrete channel set closest to the input value; The working agent determines whether to replace the original structure with the generated structure candidate according to the fitness. The quantization calculation formula for the fitness is: ; Where is the fitness of the j th pruning structure , represents the th pruned network, represents the th pruned network The corresponding weights This represents the test set. Represents the training set; The agent selection method is used to select pruning structures based on fitness-related probabilities, generating better pruning structure candidates. The quantification formula for the selection probability is as follows: ; in, For the first j The selection probability of a pruned structure This represents the maximum fitness value in the current pruning structure set; The reinitialization agent is used to reinitialize pruned structures that have not been optimized after exceeding a preset number of updates, thus preventing the search from getting trapped in local optima.

[0048] The working agent can generate new structure candidates for each pruned structure and perform structure replacement based on fitness, continuously optimizing the detection performance of pruned structures; the selection agent selects pruned structures based on fitness-related probabilities, which can generate better pruned structure candidates and improve the search efficiency of the optimal structure; the reinitialization agent can reinitialize pruned structures that have not been optimized for a long time, avoiding the search process from getting stuck in local optima and ensuring that the finally selected pruned structures have the best detection performance and adaptability.

[0049] The channel search space constraint step can narrow the combinatorial search space of model pruning and improve the search efficiency of the optimal pruning structure; by optimizing the pruning structure search step and combining the collaborative optimization algorithm to iteratively update the pruning structure set in multiple rounds, the pruning structure with the best fitness can be selected to ensure the detection accuracy of the model; through the model fine-tuning and deployment steps, the optimized lightweight model can be deployed to the embedded microcontroller after solidifying the weights, ensuring the stable operation of the model on the embedded device and realizing real-time depression detection and depression severity determination.

[0050] like Figure 7 As shown, the execution process of the variable olfactory stimulation paradigm includes: S31. In the pre-experiment phase, obtain users' subjective evaluation results on the emotional dimensions induced by different scents, and complete the labeling of scent emotional dimensions. S32. In the dimensional experiment phase, the odor-valence dimension experiment and the odor-arousal dimension experiment were performed according to the calibration results. The odor-valence dimension experiment used pleasant odors with equal probability as biased stimuli, and the odor-arousal dimension experiment used relaxing odors with equal probability as biased stimuli. Distilled water or air was used as the standard stimulus in both experiments. S33, Stimulus Presentation Phase: Deviation stimuli and standard stimuli are presented to the user in a varied and random manner, with each odor stimulus lasting for 60 seconds. S34. During the interaction and data acquisition phase, the user quickly responds by pressing a key when they perceive an odor stimulus, presses the "F" key when they perceive a deviated stimulus, and presses the "J" key when they perceive a standard stimulus. The user's key response data and real-time EEG signals are collected simultaneously.

[0051] The pre-experiment phase completes the personalized calibration of the olfactory emotional dimension, which can adapt to the differences in emotional perception of olfactory stimuli among different users and ensure the relevance of subsequent stimulation experiments. In the dimensional experiment phase, valence and arousal dimension experiments are carried out separately to comprehensively cover the emotional regulation needs of depressive disorders while reducing mutual interference between dimensions. In the stimulus presentation phase, stimuli are presented in a variable and random manner to avoid users' prediction of stimulus patterns and ensure the authenticity and effectiveness of EEG signal acquisition. In the interaction and data acquisition phase, the user's key response data and real-time EEG signals are acquired simultaneously to enrich the data dimensions for judging depressive state, while ensuring the user's attention is focused during the stimulation process and improving the effect of olfactory stimulation.

[0052] The working principle of the aforementioned control device is based on real-time acquisition and analysis of electroencephalogram (EEG) signals and closed-loop feedback control of olfactory stimulation, specifically: After the user puts on the wearable helmet body 1, the annular fixing airbag 111 of the head fixing component 11 expands by inflating and deflating, causing the silicone pad 112 to fit against the user's head, thus stably fixing the wearable helmet body 1 to the user's head. At the same time, it also causes the EEG electrodes 2 embedded inside the wearable helmet body 1 to maintain a stable fit against the user's scalp. The EEG electrodes 2 collect the user's EEG signals in real time and transmit the collected raw EEG signals to the central processing and control unit in the control mounting position 102 at the top of the wearable helmet body 1.

[0053] After receiving the raw EEG signal, the central processing and control unit first preprocesses the signal to remove environmental noise and physiological artifacts. Then, it extracts effective EEG features from the processed signal, including olfactory evoked EEG features. Through the depression state recognition algorithm preloaded on the built-in embedded microcontroller, it performs inference calculations on the effective EEG features to determine the severity of the user's current depression.

[0054] The central processing and control unit generates corresponding olfactory stimulation control commands based on the determined severity of depression using a pre-loaded adaptive adjustment algorithm, and transmits the control commands to the corresponding micro metering pump 301. Upon receiving the control commands, the micro metering pump 301 operates according to the parameters set in the commands, quantitatively extracting the mood-regulating essential oil medium stored in the corresponding detachable olfactory bottle 3 from the detachable olfactory bottle mounting position 103, thus completing the precise delivery of the essential oil medium. The device constructs its control logic based on a variable olfactory stimulation paradigm. It can first obtain the user's subjective evaluation of the emotional dimension of the odor for calibration, and then conduct experiments in the odor-valence dimension and odor-arousal dimension. During the experiments, biased stimuli and standard stimuli are presented in a variable random manner, and the user's key press response data is collected simultaneously to further improve the determination and control adaptation of the depressive state.

[0055] The essential oil medium is delivered to the metal shaping hose 4 via the micro metering pump 301, and then to the nozzle 302 at the end via the metal shaping hose 4. The nozzle 302 releases the essential oil medium into the user's nasal cavity area to stimulate the user's sense of smell and regulate the depressive state.

[0056] Throughout the olfactory stimulation process, the EEG electrode 2 continuously collects the user's real-time EEG signals and transmits them synchronously to the central processing and control unit. Based on the real-time updated EEG signals, the central processing and control unit repeatedly determines the severity of depression and adjusts the olfactory stimulation control commands accordingly. It dynamically changes the operating parameters of the micro metering pump 301 to achieve real-time adjustment of the intensity, duration, and frequency of olfactory stimulation, forming a complete closed loop of collection, identification, regulation, stimulation, and re-collection, and continuously optimizing the effect of depression regulation.

[0057] The use of the device is divided into four stages: preparation for wearing, wearing and debugging, operation and use, and post-maintenance. The specific operation procedure is as follows: The first stage is preparation for wearing. Before use, check the device's battery status and confirm that the rechargeable power module has sufficient power. According to the usage requirements, install the detachable sniff bottle 3 containing the corresponding mood-regulating essential oil medium into the sniff bottle mounting positions 103 on both sides of the wearable helmet body 1. During the installation process, the piercing protrusion at the air inlet end of the micro metering pump 301 pierces the sealing membrane of the air outlet of the detachable sniff bottle 3, and the sealing component completes the sealing of the air outlet, thus completing the installation of the sniff bottle.

[0058] The second stage is the wearing and adjustment. The wearable helmet body 1 is placed on the user's head, and its position is adjusted so that the EEG electrodes 2 correspond to the user's forehead, top of the head, temporal lobe, and occipital region. The annular fixing airbag 111 is inflated, and the expansion amount is adjusted to ensure the silicone pad 112 is evenly fitted to the user's head, the wearable helmet body 1 is stably fixed, and the EEG electrodes 2 maintain a tight fit with the user's scalp. The shape of the metal shaping hose 4 is bent and adjusted to move the nozzle 302 to the position aligned with the user's nasal cavity, completing the wearing and adjustment process.

[0059] The third stage is operation and use. Upon powering on the device, EEG electrodes 2 begin real-time acquisition of the user's EEG signals. The central processing and control unit simultaneously processes the signals and determines the severity of depression, automatically generating olfactory stimulation control commands based on the determination results. This drives the micro-quantitative pump 301 to operate, implementing corresponding olfactory stimulation modulation. During operation, the user can manually adjust the positions of the metal-shaped tubing 4 and the nozzle 302 based on their experience. The central processing and control unit can also switch the working and standby olfactory bottle groups and change the type of essential oil medium. The device automatically acquires continuous EEG signals and dynamically adjusts the control parameters throughout operation, requiring no additional manual intervention. When conducting experiments with varying olfactory stimulation paradigms, the user must respond by pressing a corresponding button upon experiencing odor stimulation. The device simultaneously acquires button data and EEG signals to optimize subsequent control strategies.

[0060] The fourth stage is the final maintenance. After use, turn off the device power, deflate the annular fixed airbag 111, and remove the wearable helmet body 1. After the essential oil medium in the detachable sniff bottle 3 is used up, it can be removed from the sniff bottle mounting position 103 and replaced with a new detachable sniff bottle 3. Regularly clean the inside of the wearable helmet body 1, the silicone pad 112, the EEG electrodes 2, and the nozzle 302 to ensure the hygiene and safety of the device.

[0061] Therefore, this invention employs the aforementioned wearable olfactory stimulation modulation device and method, using an integrated helmet as a carrier to integrate EEG acquisition, embedded local processing, dual-channel quantitative odor generation and directional atomization delivery units. Through the COPruner lightweight model, it achieves real-time local identification of depressive states using EEG signals. Combined with an adaptive algorithm, it generates stimulation commands to drive a micro-quantitative pump and a shapeable flexible nozzle to complete precise, targeted, and leak-free olfactory stimulation. Furthermore, it relies on real-time EEG feedback to form a closed-loop adaptive modulation. Simultaneously, an adaptive airbag fixation structure ensures stable and comfortable wear. Overall, it solves the problems of poor portability, recognition delay, uncontrollable stimulation, and lack of dynamic adjustment in existing devices, achieving daily, non-invasive, and personalized auxiliary intervention for depressive disorders.

[0062] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method of olfactory stimulation modulation by wearable device, characterized in that, Includes the following steps: S1. The raw EEG signal is preprocessed to remove interference artifacts and environmental noise, and effective EEG features are extracted. A lightweight depression recognition model based on the feature map pruning method COPruner is used to infer and calculate the effective EEG features to determine the severity of the user's current depressive disorder. S2. Based on the severity of depression, an adaptive adjustment algorithm is used to generate olfactory stimulation control instructions, using EEG depression biomarkers as the regulation target to determine the odor type, delivery intensity, stimulation duration and stimulation frequency.

2. The wearable olfactory stimulation modulation method according to claim 1, characterized in that, The preprocessing operations of the raw EEG signal in S1 include baseline correction, power frequency filtering, removal of electrooculography artifacts and removal of electromyography artifacts. The preprocessed EEG signal is then subjected to feature extraction to obtain EEG features related to the severity of depression. The EEG feature indicators include the frequency domain features, time domain features and nonlinear features of the EEG signal.

3. The wearable olfactory stimulation modulation method according to claim 1, characterized in that, The methods for constructing a mild depression identification model in S1 include: S11. Channel Search Space Constraint: For the ResNet 3-layer basic model, the number of channels retained in each convolutional layer is limited to a preset discrete set. Within this framework, the combinatorial search space is narrowed down through constraints; among which, For the original convolutional layer j Number of channels in the layer Reserve the maximum percentage for the preset channels; S12, Pruning Structure Optimization Search: Initialize the pruning structure set, pre-train the base model weights, take maximizing classification accuracy as the optimization goal, combine the feature map pruning method COPruner based on group collaboration optimization, iteratively update the pruning structure set, and select the pruning structure with the best fitness. S13. Model fine-tuning and deployment: Fine-tune the lightweight model corresponding to the optimal pruning structure, solidify the model weights, and deploy it to the embedded microcontroller for real-time depression detection and severity determination.

4. The wearable olfactory stimulation modulation method according to claim 3, characterized in that, The quantitative formula for the optimization objective is: ； in, This is the network structure of the pruned model. This is the pruned CNN model. For the pruned model on the training set Weights for training or fine-tuning For the model on the test set The accuracy rate of depression detection on the device; Introducing channel constraints, the optimization objective is constrained as follows: ； in, For the model after pruning i Number of feature map channels in the layer L This represents the total number of convolutional layers in the CNN model.

5. The wearable olfactory stimulation modulation method according to claim 3, characterized in that, In the feature map pruning method COPruner, which is based on collaborative optimization, the pruning structure set is iteratively updated through working agents, agent selection, and agent re-initialization to select the pruning structure with the best fitness. The working agent is used to generate new structure candidates for each pruning structure. The quantization generation formula for the new structure candidates is as follows: ； in, For the generated new structure candidate i Number of channels in the layer The original pruning structure i Number of channels in the layer r A random number in the range of 0-1. The current optimal pruning structure is the i Number of channels in the layer A rounding function that returns the value within a preset set of discrete channels that is closest to the input value; The working agent determines whether to replace the original structure with the generated structure candidate based on fitness. The formula for quantifying fitness is as follows: ； in, For the first j A pruning structure Adaptability, Indicates the first A pruned network Indicates the first A pruning network The corresponding weights This represents the test set. Represents the training set; The agent selection method is used to select pruning structures based on fitness-related probabilities, generating better pruning structure candidates. The quantification formula for the selection probability is as follows: ； in, For the first j The selection probability of a pruned structure This represents the maximum fitness value in the current pruning structure set; The reinitialization agent is used to reinitialize pruned structures that have not been optimized after exceeding a preset number of updates, thus preventing the search from getting trapped in local optima.

6. A wearable olfactory stimulation modulation device, based on the wearable olfactory stimulation modulation method according to any one of claims 1-5, characterized in that, The device includes a wearable helmet, an EEG signal acquisition unit, a central processing and control unit, an odor generation unit, and an odor-directed delivery unit. The wearable helmet has a head-fixing component inside, several ventilation holes on its outer surface, a control mounting position on its top, and snorkel mounting positions symmetrically arranged on both sides. The EEG signal acquisition unit includes multiple sets of EEG electrodes embedded inside the helmet, which maintain stable contact with the user's scalp via the head-fixing component. A central processing and control unit is installed in the control mounting position, and contains an embedded microcontroller. The odor generation unit includes detachable snorkels and corresponding micro-metering pumps. The detachable snorkels include a working snorkel group and a spare snorkel group. The detachable snorkels are installed in the snorkel mounting positions. The inlet of the micro-metering pump is sealed to the outlet of the corresponding detachable snorkel, and the controlled end of the micro-metering pump is electrically connected to the control output of the central processing and control unit. The inlet of the odor-directed delivery unit is connected to the outlet of the micro-metering pump, and the outlet of the odor-directed delivery unit has a nozzle with adjustable spatial position, positioned towards the user's nasal cavity.

7. A wearable olfactory stimulation modulation device according to claim 6, characterized in that, The head fixation component includes an annular fixation airbag adapted to the user's head circumference and silicone pads fixedly installed on the inner annular surface of the annular fixation airbag. The annular fixation airbag is arranged around the inner edge of the wearable helmet body, and the silicone pads are evenly distributed on the inner annular surface of the annular fixation airbag.

8. A wearable olfactory stimulation modulation device according to claim 6, characterized in that, The EEG electrodes are placed on both sides of the head, the forehead, the top of the head, and the occipital region of the wearable helmet. The rear end of each EEG electrode is equipped with an elastic buffer.

9. A wearable olfactory stimulation modulation device according to claim 6, characterized in that, The detachable sniff bottle has a sealing membrane at the air outlet, and the air inlet of the micro metering pump is equipped with a puncture protrusion and a seal. The detachable sniff bottle contains essential oil media, which includes valence-inducing and arousal-inducing scents.

10. A wearable olfactory stimulation modulation device according to claim 6, characterized in that, The odor delivery unit includes a metal-shaped flexible tube, one end of which is sealed and connected to the outlet of a micro metering pump, and the other end of which is fixedly connected to a nozzle.

Images