Method for analyzing optimal single-time drinking amount of white spirit based on electroencephalogram signals

Abstract

The invention provides an electroencephalogram signal-based method for analyzing the optimal single-time drinking amount of white spirit. The electroencephalogram signal-based method comprises the following steps: (a) collecting electroencephalogram signals of a subject in the drinking process of drinking white spirit with different preset drinking amounts; (b) performing multi-dimensional analysis on the electroencephalogram signals, including spectral analysis, time-frequency analysis and event-related spectrum disturbance analysis, the spectral analysis including difference analysis of power spectrum values and brain region specificity analysis; and (c) determining the drinking amount corresponding to the optimal neural activity state as the optimal drinking amount based on the analysis result of the step (b). According to the method provided by the invention, the continuous neural activity change after drinking is recorded, and the brain neural activity after drinking is comprehensively monitored and interpreted in combination with a multi-dimensional electroencephalogram data analysis method, so that a reference can be provided for the single drinking amount when a consumer drinks the white spirit, and the drinking quality of the consumer is improved. And a more reliable neuroscience basis is provided for wine product design and consumer experience optimization, and further development of the wine product and brain nerve activity field is promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of electroencephalogram (EEG) signals, and in particular to an analysis method for determining an optimal single consumption of liquor based on EEG signals. Background Art

[0002] With the continuous development of the beverage and food industries, consumers' demands for a drinking experience go beyond taste and flavor. The psychological and physiological effects of drinking, particularly the potential impact on brain neural activity, are attracting widespread attention. Therefore, how to scientifically and objectively evaluate the impact of different sample sizes on neural activity, such as cognitive function, emotional regulation, and behavioral responses, has become a pressing technical challenge in the food and beverage industry.

[0003] Traditional methods for evaluating the comfort of drinking products rely on subjective questionnaires and scales. While these methods reflect consumer perceptions to a certain extent, they are often subjective due to individual differences and subjective tendencies. In recent years, electroencephalogram (EEG) technology has become an important tool for studying brain activity due to its immediacy, non-invasiveness, and high sensitivity. However, existing EEG analysis methods mostly focus on static or short-term neural responses.

[0004] CN112137614A proposes a motion feature recognition method related to stereoscopic video comfort based on EEG signals. This technology combines stereoscopic video (3D display technology) and EEG signals to objectively evaluate the viewer's comfort through real-time changes in brain neural activity. However, this patent mainly relies on SVM classifiers and CSP feature extraction methods for classification. Although effective in specific tasks, its method is limited to binary classification scenarios and has difficulty revealing complex neural activities and the brain's response to dynamic stimuli. It also does not fully utilize spectrum or rhythm wave analysis to obtain more in-depth neural activity information. CN111798974A proposes an evaluation method combining ANT tasks and EEG / ERP potential analysis for the assessment of the comfort level after drinking alcoholic products. This method objectively evaluates the comfort level after drinking different alcoholic products through neuroelectrophysiological signals. However, the patent adopts a fixed experimental design and most of the participants are professional wine tasters, which may have limited applicability to a wider population. In addition, the study focuses on the assessment of comfort level after drinking and does not involve more complex EEG response patterns (such as event-related spectral perturbation, i.e., ERSP analysis). Therefore, it may be insufficient in capturing the dynamic impact of alcoholic products on EEG signals.

[0005] Existing technologies demonstrate the widespread application of EEG technology in human comfort assessment, but they primarily focus on static or short-term neural responses and fail to fully capture the ongoing changes in neural activity during and after consumption, particularly of alcoholic beverages. Furthermore, many methods rely solely on frequency domain analysis or simple ERPs analysis, failing to fully integrate more complex EEG analysis techniques. These limitations, such as the single nature of the analytical methods and sample applicability, limit their further application and development. Summary of the Invention

[0006] In view of the above problems in the prior art, the first aspect of the present invention provides a method for determining the optimal drinking amount based on EEG signal analysis, which is characterized by comprising the following steps:

[0007] (a) Collecting EEG signals of subjects during the drinking process of liquor at different preset drinking volumes;

[0008] (b) Multi-dimensional analysis of EEG signals, including:

[0009] Spectral analysis, time-frequency analysis and event-related spectral perturbation analysis,

[0010] The spectrum analysis includes difference analysis of power spectrum values ​​and brain region specificity analysis;

[0011] (c) Based on the analysis results of step (b), determining the drinking amount corresponding to the optimal neural activity state as the optimal drinking amount.

[0012] In some embodiments, in step (a), the screening conditions for the subject include: age within a preset range, no history of alcohol dependence, and no intake of substances that affect neural activity before the test, and the preset drinking amount includes at least three gradient doses.

[0013] In some embodiments, the EEG signal acquisition process in step (a) includes the following steps according to the subject's drinking process: acquisition using a full-scalp electrode array with a sampling frequency of 1000 Hz, focusing on monitoring the frontal, parietal, and occipital lobe channels; signal recording starts from the sample entrance and continues for 5 seconds before swallowing and 2 minutes to 2.5 minutes after swallowing, and 10 parallel data of 120 seconds to 150 seconds are collected for each drinking amount; after a single test, the subject rests for 5 minutes to restore the baseline.

[0014] In some embodiments, step (b) further comprises establishing a taste EEG dataset using the acquired EEG signals, and preprocessing the taste EEG dataset, wherein the preprocessing process comprises removing useless electrodes, re-referencing, filtering, segmentation and baseline correction, interpolating bad leads and removing bad segments, performing ICA, and removing noise components;

[0015] In some embodiments, the EEG data set is segmented into 40-50s and 50-60s segments to create 800 samples, and each sample is sequentially subjected to a 30 Hz low-pass filter and a 1 Hz high-pass filter to achieve noise reduction of the EEG signal.

[0016] In some embodiments, the power spectrum value difference analysis in step (b) is to calculate the changes in power spectrum values ​​of rhythmic waves under different drinking amounts, including: calculating the power spectrum density of the rhythmic waves by the Pwelch method, performing Welch's variance test on the power value, and screening out rhythmic waves with significant differences in power spectrum values ​​P<0.5 between different drinking amounts; the rhythmic waves can be divided into 0-4Hz delta waves, 4-8Hz theta waves, 8-12Hz alpha waves, 12-20Hz beta waves and 20-30Hz gamma waves according to the frequency of the electroencephalogram.

[0017] In some embodiments, the brain region specific analysis is to evaluate the differences in the distribution of rhythmic wave power spectrum values ​​in at least one brain region, including comparing the changes in the power spectrum values ​​of delta waves, theta waves, alpha waves and gamma waves in different brain regions under different drinking amounts; in some embodiments, the brain regions include the central region, the prefrontal region, the left temporal lobe region, the parietal lobe region and the right temporal lobe region.

[0018] In some embodiments, the time-frequency analysis includes using a short-time Fourier transform method to analyze the energy distribution characteristics of the EEG signal in time and frequency, which can be used for event-related spectral disturbance analysis; the time is 40-50s and 50-60s, and the frequency is 0-4Hz, 4-8Hz, 8-12Hz, 12-20Hz and 20-30Hz.

[0019] In some embodiments, the event-related spectral perturbation analysis includes selecting representative channels from the frontal lobe, central region, parietal lobe, and occipital lobe to extract and analyze the 0-40 Hz data response results, using event-related spectral perturbation to process the experimental data, and analyzing the correlation between the response time delay of the rhythmic wave and the amount of drinking.

[0020] In some embodiments, the optimal neural activity state includes at least the following state: in the power spectrum value specific analysis, the power spectrum value of the delta band is between 474.36-536.95 and the power spectrum value of the theta band is between 32.08-36.12.

[0021] In some embodiments, step (c) further includes establishing a fitting model based on the multidimensional analysis results, and the fitting model can output a recommended drinking amount; in some embodiments, y = -0.943 + 0.009 × PSD (δ) + 0.01 × PSD (θ) - 0.253 × PSD (γ) - 0.042 × PSD (α) + 0.113 × PSD (β), in which PSD (δ), PSD (θ), PSD (γ), PSD (α) and PSD (β) are the power spectral density values ​​of δ, θ, γ, α and β waves, respectively, and y represents the recommended optimal drinking amount.

[0022] In some embodiments, step (c) also includes verifying the fitting model, specifically collecting EEG signals from different subjects, predicting the optimal drinking amount based on the power spectrum value and the corresponding fitting model, and comparing it with the actual drinking amount, analyzing the difference between the predicted value and the actual value, and judging the accuracy of the prediction model.

[0023] The second aspect of the present invention provides an optimal drinking amount recommendation system, which is characterized by including: an EEG signal acquisition module, a multidimensional analysis module, and a drinking amount calculation module; the EEG signal acquisition module is used to obtain multi-channel EEG data of the subject during the drinking process; the multidimensional analysis module performs the multidimensional analysis described in any one of the first aspects; and the drinking amount calculation module outputs a recommended drinking amount based on the analysis results.

[0024] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the steps of any one of the methods described in the first aspect.

[0025] The present invention monitors and records the EEG response to drinking different sample amounts for a continuous period of time through rigorous experimental design and optimized experimental paradigm. It adopts multiple analysis methods, such as difference analysis, spectrum analysis, analysis of different rhythmic waves and ERSP analysis, to evaluate the brain response under different drinking amounts and derive the recommended optimal drinking amount. This data can provide a reference for consumers' single drinking amount when drinking white wine. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is an experimental paradigm for collecting EEG signals of subjects at various drinking levels;

[0027] Figure 2 The changes in delta waves in different brain regions under different drinking amounts;

[0028] Figure 3 The changes in theta waves in different brain regions under different drinking amounts;

[0029] Figure 4The changes in alpha waves in different brain regions under different drinking amounts;

[0030] Figure 5 The changes of gamma waves in different brain regions under different drinking amounts;

[0031] Figure 6 Figure 3. EEG spectrum analysis results for different sample volumes. A, B, C, D, and E represent the EEG spectrum analysis results for 0.75 mL, 6.00 mL, 9.00 mL, 12.00 mL, and 15.00 mL, respectively.

[0032] Figure 7 The graph shows the time-frequency analysis results of EEG under different sample sizes;

[0033] Figure 8 The ERSP analysis results on the four channels of Fz, Cz, Pz and Oz within the time period of 50-90s under different drinking amounts. DETAILED DESCRIPTION

[0034] The following is a detailed description of the technical solution of the present invention, which does not limit the scope of protection of the present invention. Non-essential modifications and adjustments made by others based on the concept of the present invention still fall within the scope of protection of the present invention.

[0035] Electroencephalogram (EEG): An electrophysiological method for recording brain activity. EEG is formed by the summation of postsynaptic potentials generated synchronously by numerous neurons during brain activity. It records changes in electrical activity during brain activity and is a comprehensive reflection of the electrophysiological activity of brain nerve cells in the cerebral cortex or on the surface of the scalp.

[0036] Rhythmic waves: According to the EEG of different bands, it can be divided into 5 types of rhythmic waves, namely δ (0-4Hz), θ (4-8Hz), α (8-12Hz), β (12-20Hz) and γ (20-30Hz); among them, δ wave and θ wave are the main manifestations of electrical activity when the cerebral cortex is in an inhibited state. When δ wave is the dominant brain wave, people are usually in a deep sleep and unconscious state. When θ wave is the dominant brain wave, it is usually when normal people are sleeping, with interrupted consciousness and deep relaxation of the body. In addition, θ wave has a very direct relationship with the limbic system of the brain, which is very helpful for triggering deep memory and strengthening long-term memory (LTP). Alpha wave is closely related to the brain. Alpha waves are closely related to the brain's relaxed state, promoting inspiration, accelerating information collection, and enhancing memory. They are the optimal brainwaves for learning and thinking. They typically appear when awake, quiet, and with eyes closed. When eyes are open, thinking, or receiving other stimuli, alpha waves disappear. Beta waves are the primary manifestation of electrical activity when the cerebral cortex is tense and excited. They play a positive role in coordination and communication and are the most common high-frequency waves experienced during wakefulness. When beta wave activity increases, the brain limits the number of alpha waves. Gamma waves are the primary manifestation of electrical activity when the cerebral cortex is tense and excited. They play a positive role in coordination and communication and are closely related to consciousness, memory, and learning ability. They are typically active during complex cognitive tasks or inspiration.

[0037] Short-Time Fourier Transform (STFT): A time-frequency analysis method used to analyze non-stationary signals. It slides a time window over the signal and performs a Fourier transform on the signal within each window, thereby obtaining the frequency information of the signal at different time points.

[0038] Example 1

[0039] Experimental subjects: There were 5 subjects participating in the experiment, including 3 males and 2 females. The subjects were all between 24 and 30 years old, did not smoke, had no history of illness, had a history of drinking for half a year or more, and had no alcohol dependence behavior. The subjects were not allowed to drink, smoke, drink coffee, or take drugs 24 hours before the experiment.

[0040] Experimental evaluation samples: Maotai-flavor liquor. The quality of the experimental samples was screened by national-level wine tasters, and Moutai liquor with a mellowness level of 5 (0 to 9 points) was selected.

[0041] Preset drinking volume setting: The experiment selected 5 different sample drinking volumes, namely 0.75mL, 6.00mL, 9.00mL, 12.00mL and 15.00mL. The specific values ​​of each drinking volume were set based on sensory drinking volume, daily drinking volume and experimental requirements.

[0042] The following experiments were conducted in an independent and quiet environment. The specific implementation plan is as follows:

[0043] 1. Collect EEG signals of subjects during the drinking process of liquor at different preset drinking amounts

[0044] The subjects drank different amounts of liquor, including a pre-drinking preparation, drinking, and a rest period. The pre-drinking preparation involved rinsing the mouth 30 seconds before drinking. The drinking process involved holding the sample in the mouth for 5 seconds before swallowing, followed by a 2-minute savoring period. The post-drinking rest period consisted of a 5-minute rest period. All tests were conducted on the same group of subjects, with each subject undergoing five different drinking sequences.

[0045] The EEG signals of the subjects were collected during the drinking process. The EEG signal recording covered the entire scalp electrode array, focusing on monitoring specific brain areas, such as the frontal lobe, parietal lobe and occipital lobe, and collecting the brain's neural response data under different drinking amounts.

[0046] The EEG signal acquisition process is as follows Figure 1 The experimental paradigm shown here begins recording from the moment the sample is ingested and continues for 2.5 minutes. This is the EEG signal corresponding to a single drinking volume. After recording, the subjects underwent a 5-minute rest period to ensure that EEG activity returned to baseline. During this rest period, the subjects remained quiet and minimized external distractions to prevent environmental interference with the EEG signal. EEG signals were collected using a 1000Hz EEG signal acquisition system. Taste stimulation EEG data were collected from five subjects. For each subject, five taste EEG data sets were collected for different drinking volumes, with each taste stimulus consisting of ten parallel EEG data sets of 150 seconds each.

[0047] 2. Multi-dimensional analysis of collected EEG signals

[0048] 2.1 Use the acquired EEG data to establish a taste EEG dataset, and perform preprocessing such as filtering and ICA in sequence.

[0049] Sample preprocessing included removing unused electrodes, re-referencing, filtering, segmentation and baseline correction, interpolating bad leads and removing bad segments, performing ICA, and removing noise. When creating the EEG dataset, the collected EEG data were segmented into 40-50s and 50-60s segments, creating 800 samples. Each sample was then subjected to a 30Hz low-pass filter and a 1Hz high-pass filter to reduce EEG signal noise.

[0050] 2.2 Perform time-frequency analysis and spectrum analysis on the preprocessed data to distinguish the differences.

[0051] 2.2.1 Spectral Analysis: Fourier transform was performed on the signals of multiple subjects on different EEG channels. The power spectral density (PSD) of each segment was calculated using the Pwelch method. The power mean and peak frequency of each frequency band were extracted from the power spectral density, and the frequency bands were divided into delta, theta, alpha, beta, and gamma waves.

[0052] ① The power spectrum values ​​of different frequency bands (δ, θ, α, β, and γ) at different drinking amounts were calculated, and Welch's variance test was performed on the power spectrum values ​​of different frequency bands at five different drinking amounts.

[0053] ②Analyze the changes in delta, theta, alpha, and gamma waves in different brain regions under different drinking amounts. Analyze the power spectrum values ​​of the brain C (central area), F (frontal area), LT (left temporal lobe area), PO (parietal lobe), and RT (right temporal lobe area);

[0054] ③Calculate the spectral amplitude and spectral power, and perform group-level analysis and visualization.

[0055] 2.2.2 Time-Frequency Analysis: Short-time Fourier transform (STFT) analysis was primarily used to analyze the time-frequency characteristics of EEG signals. Time-frequency maps at different electrodes, baseline-corrected time-frequency maps, power changes across different brain regions, and scalp topography were displayed. Region selection and ROI value extraction were performed at time intervals of 40-50s and 50-60s, and within the frequency ranges of 0-4Hz, 4-8Hz, 8-12Hz, 12-20Hz, and 20-30Hz.

[0056] 2.2.3 Event-related spectral perturbation (ERSP) analysis: Representative channels (Fz, Cz, Pz, and Oz) from the frontal, central, parietal, and occipital regions were selected for data extraction and analysis of the 0-40 Hz response results. Event-related spectral perturbation was used to process the experimental data and conduct a comparative analysis.

[0057] 3. Based on the analysis results of EEG signals, the drinking amount corresponding to the optimal neural activity state is determined as the optimal drinking amount.

[0058] 3.1 Spectrum Analysis Results

[0059] Table 1 shows the results of a differential analysis of the power spectrum values ​​corresponding to different brain waves at different drinking levels. A variance analysis of the corresponding brain wave power spectra was performed based on the variance test. The results showed that the power spectrum values ​​corresponding to delta, theta, alpha, and gamma brain waves were significantly different at different drinking levels (P < 0.5), suggesting that different drinking levels may affect these four rhythmic waves.

[0060] Table 1 Analysis of power spectrum differences of brain waves corresponding to different drinking amounts

[0061]

[0062]

[0063] In order to explore the brain activity corresponding to different drinking amounts, we further analyzed the changes in δ, θ, α, and γ waves in different brain areas under different drinking amounts. The results are as follows: Figures 2 to 5 shown.

[0064] When the drinking volume was 6.00 mL, the α wave power spectrum values ​​of C (central area), F (frontal area), LT (left temporal lobe area), PO (parietal lobe) and RT (right temporal lobe area) were the highest, but the δ, θ and γ wave response values ​​were also very high, which may have an impact on the brain function of these areas.

[0065] When the drinking volume was 9.00 mL, the alpha wave power spectrum values ​​in most brain regions were lower, but the delta, theta, and gamma wave response values ​​were also lower than those at other drinking volumes. Therefore, the corresponding areas would have relatively little impact on brain function.

[0066] When the drinking volume was 15.00 mL, the delta, theta, and gamma wave responses in the C, F, LT, and RT regions all reached their highest values, potentially indicating a significant impact on brain function in these areas. This may include increased sensitivity to physical sensations. Combined with wavelength analysis, it can be speculated that the peak wavelength response values ​​in different brain regions may reflect a state of relaxation and recovery. Based on these findings, both 6.00 mL and 9.00 mL are likely optimal drinking volumes.

[0067] The collected EEG data were further analyzed by spectrum analysis. Figure 6 As can be seen, at both 6.00 mL and 9.00 mL of alcohol, alpha waves showed strong responses in the frontal, central, parietal, and occipital regions. The alpha wave response at 9.00 mL was superior to that at 6.00 mL. Alpha waves are closely associated with a state of brain relaxation, and the alpha wave response showed significant differences between different drinking volumes, with the highest alpha wave response observed at moderate drinking volumes. Theta wave responses at both 6.00 mL and 9.00 mL showed strong responses across the entire brain at 9.00 mL, while at 6.00 mL, theta waves showed strong responses across most brain regions, as shown by topographic maps. Theta waves are closely associated with the regulation of emotion, memory, and attention. Therefore, the 9.00 mL drink indicates a relatively relaxed brain state and better regulation of these functions. Beta waves also showed responses in the occipital lobe, but the differences were relatively minor.

[0068] Therefore, from the above results, it can be seen that when the alcohol consumption is 9.00mL, the brain's alpha waves show obvious differences in different brain regions, showing the highest alpha wave response. At this time, the brain is in a relaxed state and has a better ability to regulate emotions, memory and attention.

[0069] 3.2 Time-frequency analysis results

[0070] The time-frequency analysis results are as follows Figure 7 As shown in the figure, the analysis results show that within 150 seconds after drinking, the subjects' brains all showed certain responses, and the response effects of delta waves, theta waves, alpha waves, and beta waves were different at different times. Within 0-10 seconds, the responses of delta waves and theta waves were better when the drinking volume was 9.00mL, 12.00mL, and 15.00mL than those at other drinking volumes. The alpha wave showed a better response immediately after the entrance, and the difference in response gradually weakened in the later stages of drinking. This may be related to the fact that the immediate stimulation after the entrance of white wine is more palatable than the stimulation at other drinking volumes. At this time, 15.00mL was also accompanied by a beta wave response, indicating that in addition to the relaxation state, this drinking volume also caused greater psychological activity or emotional fluctuations, which has a certain impact on the palatability of the drinking volume. When the alcohol consumption was 0.75mL and 6.00mL, the alpha wave had a continuous response within the acquisition time. As time went on, the difference was not significant. This also indirectly indicates that this amount of drinking has an impact on the brain's response, but the impact may be small, and it does not reflect the changes in the oral perception of liquor for a period of time after drinking.

[0071] Therefore, combined with the above conclusions, the brain responses under drinking volumes of 9.00mL and 12.00mL both showed good differentiation effects, but the drinking volume of 9.00mL was better than 12.00mL and was more suitable as the optimal drinking volume.

[0072] 3.3 Results of event-related spectrum disturbance analysis

[0073] Combining artificial sensory and EEG data analysis, the feedback after drinking is mainly related to the characteristics of the aftertaste. Therefore, based on artificial sensory, we intercepted 50-90s of EEG data, selected the representative channels of the four regions F, C, P and O, namely Fz, Cz, Pz and Oz, to extract and analyze the 0-40Hz data response results, and used the ERSP method to process the experimental data and conduct comparative analysis. The analysis results are as follows: Figure 8 shown.

[0074] The distribution of response times reveals that the response time to the aftertaste of a drink is delayed as the amount consumed increases. For example, in the Fz channel, when the amount consumed was 0.75mL, 6.00mL, 9.00mL, 12.00mL, or 15.00mL, the brain responses were concentrated within five distinct time periods: 50-60s, 50-65s, 60-65s, 65-70s, and 75-85s, respectively. This suggests that the response to the aftertaste of a drink shifts later with increasing single-drink sample volume.

[0075] The response results from the four channels at different sample sizes show that the responses in the δ, θ, and α bands are superior to those in other bands within this timeframe. Furthermore, as the amount consumed increases, the responses in the β and γ bands are superior to those in the lower amounts consumed. Based on the current analysis, it is clear that consuming baijiu produces a persistent brain response, and that larger single sample sizes and longer durations of response contribute to a greater impact on emotional perception, particularly comfort.

[0076] 3.4 Conclusion

[0077] The above analysis shows that different drinking amounts significantly affect brain neural activity. 9.00 mL is the recommended optimal drinking amount for the subjects. At this drinking amount, the subjects' brain alpha waves showed significant differences across different brain regions, with the highest alpha wave response indicating a relaxed brain with better regulation of emotions, memory, and attention. This also better reflects changes in oral perception of liquor after a period of drinking.

[0078] The five subjects and five drinking amounts used in the embodiment are only examples to verify the feasibility. Those skilled in the art can optimize the model accuracy by expanding the sample size and increasing the drinking amount gradient.

[0079] Example 2

[0080] This example established an optimal drinking amount model and verified its accuracy.

[0081] 1. Subjects: A total of 5 subjects participated in the validation experiment. The subjects were all between 24 and 30 years old, non-smokers, with no history of illness, a history of drinking for six months or more, and no alcohol dependence. The subjects were not allowed to drink alcohol, smoke, drink coffee, or take medications within 24 hours before the experiment.

[0082] 2. Sample evaluation and EEG signal collection process: Same as Example 1, except that the drinking amount is determined by the subject.

[0083] 3. EEG signal data processing process: The processing method described in Example 1 includes performing Fourier transform on the signals of multiple subjects on different EEG channels, using the Pwelch method to calculate the power spectral density (PSD) of each segment, and extracting the power mean and peak frequency of each frequency band from the power spectral density, dividing the frequency bands into δ, θ, α, β and γ waves, and performing the same processing on the collected data as in Example 1.

[0084] 4. Establishment of fitting model: The power spectral density values ​​of δ, θ, α, β and γ waves were optimized based on the Lasso regression model to obtain the following fitting equation:

[0085] y=-0.943+0.009×PSD(δ)+0.01×PSD(θ)-0.253×PSD(γ)-0.042×PSD(α)+0.113×PSD(β),

[0086] Wherein, PSD(δ), PSD(θ), PSD(γ), PSD(α), and PSD(β) are the power spectral density values ​​of δ, θ, γ, α, and β waves, respectively, and y is the amount of drinking.

[0087] 5. Verification of Fitting Model Accuracy: Subjects determined their own drinking volume and informed the experimenter after data analysis was complete. This volume served as validation data y. The Pwelch method was used to calculate the power spectral density (PSD) for each segment. The mean power and peak frequency of each frequency band were extracted from the PSD, which was then categorized as delta, theta, alpha, beta, and gamma waves. These values ​​were substituted into the model calculation formula to obtain the drinking volume y1. This was then compared with y and the error rate analyzed. The final validation accuracy of the fitting model was 90%.

[0088] Comparative Example 1

[0089] The experimental process of this comparative example is exactly the same as that of Example 2. The only difference is the establishment of the fitting model. This comparative example establishes the optimal drinking amount model based on the ridge regression model, and obtains the following fitting equation comparison model:

[0090] Drinking amount = -0.226 + 0.008 × PSD (δ) + 0.024 × PSD (θ) - 0.032 × PSD (α) + 0.059 × PSD (β) + 0.39 × PSD (γ), where PSD (δ), PSD (θ), PSD (γ), PSD (α) and PSD (β) are the power spectral density values ​​of δ, θ, γ, α and β waves, respectively. The accuracy was verified using the method of Example 2, and it was found that the accuracy of the fitting model of this comparative example was only 73%.

Claims

1. A method for determining the optimal drinking amount based on EEG signal analysis, characterized in that: The following steps are involved: (a) Collecting EEG signals of subjects during the drinking process of liquor at different preset drinking volumes; (b) Multi-dimensional analysis of EEG signals, including: Spectral analysis, time-frequency analysis and event-related spectral perturbation analysis, The spectrum analysis includes difference analysis of power spectrum values ​​and brain region specificity analysis; (c) Based on the analysis results of step (b), the drinking amount corresponding to the optimal neural activity state is determined as the optimal drinking amount.

2. The method according to claim 1, wherein In step (a), the screening conditions for the subjects include: age within a preset range, no history of alcohol dependence, and no intake of substances that affect neural activity before the test, and the preset drinking amount includes at least three gradient doses.

3. The method according to claim 1, wherein The EEG signal acquisition process in step (a) includes: using a full-scalp electrode array with a sampling frequency of 1000 Hz to focus on monitoring the frontal, parietal, and occipital lobe channels; signal recording starts from the sample entrance and continues for 5 seconds before swallowing and 2 minutes to 2.5 minutes after swallowing; 8-12 parallel data of 120 seconds to 150 seconds are collected for each drinking amount; Preferably, after collecting a single EEG signal, the subject rests for 5 minutes to restore the baseline.

4. The method according to claim 1, wherein The step (b) also includes establishing a taste EEG dataset using the acquired EEG signals and preprocessing the taste EEG dataset, wherein the preprocessing process includes removing useless electrodes, re-referencing, filtering, segmentation and baseline correction, interpolating bad leads and removing bad segments, ICA, and removing noise components.

5. The method according to claim 1, wherein The power spectrum value difference analysis in step (b) is to calculate the changes in the power spectrum values ​​of the rhythmic waves under different drinking amounts, including: calculating the power spectrum density of the rhythmic waves by the Pwelch method, performing Welch's variance test on the power values, and screening out the rhythmic waves with significant differences in power spectrum values ​​P < 0.5 between different drinking amounts; Preferably, the rhythmic waves can be divided into 0-4 Hz delta waves, 4-8 Hz theta waves, 8-12 Hz alpha waves, 12-20 Hz beta waves and 20-30 Hz gamma waves according to the frequency of the electroencephalogram; Preferably, the brain region-specific analysis in step (b) is to evaluate the distribution difference of rhythmic wave power spectrum values ​​in at least one brain region, including comparing the power distribution differences of delta wave, theta wave, alpha wave, and gamma wave power spectrum values ​​in different brain regions under different drinking amounts, and the brain regions include the central region, prefrontal region, left temporal lobe region, parietal lobe region, and right temporal lobe region; Preferably, the time-frequency analysis in step (b) comprises: using a short-time Fourier transform method to analyze the energy distribution characteristics of the EEG signal at a time of 40-50s and 50-60s and a frequency of 0-4Hz, 4-8Hz, 8-12Hz, 12-20Hz and 20-30Hz; Preferably, the event-related spectral perturbation analysis in step (b) includes selecting representative channels of the four regions of the frontal lobe, central area, parietal lobe and occipital lobe to extract and analyze the 0-40Hz data response results, using event-related spectral perturbation to process the experimental data, and analyzing the correlation between the response time delay of the rhythmic wave and the amount of drinking.

6. The method according to claim 1, wherein The optimal neural activity state at least includes the following state: in the power spectrum value specificity analysis, the power spectrum value of the delta band is between 474.36-536.95 and the power spectrum value of the theta band is between 32.08-36.

12.

7. The method according to claim 1, wherein The step (c) further includes establishing a fitting model based on the multi-dimensional analysis results, wherein the fitting model can output a recommended drinking amount, and the fitting model is: y=-0.943+0.009×PSD(δ)+0.01×PSD(θ)-0.253×PSD(γ)-0.042×PSD(α)+0.113×PSD(β), In the formula, PSD(δ), PSD(θ), PSD(γ), PSD(α) and PSD(β) are the power spectral density values ​​of δ, θ, γ, α and β waves, respectively, and y represents the recommended optimal drinking amount.

8. The method according to claim 7, wherein The step (c) also includes verifying the fitting model, specifically collecting EEG signals from different subjects, predicting the optimal drinking amount based on the power spectrum value and the corresponding fitting model, and comparing it with the actual drinking amount, analyzing the difference between the predicted value and the actual value, and judging the accuracy of the prediction model.

9. An optimal drinking amount recommendation system, characterized in that: include: EEG signal acquisition module, multi-dimensional analysis module, and drinking amount calculation module; The EEG signal acquisition module is used to obtain multi-channel EEG data of the subject during the drinking process; The multi-dimensional analysis module performs the method according to any one of claims 1 to 8; The drinking amount calculation module outputs a recommended drinking amount based on the analysis result.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images