Study of EEG and Nasal Stimulus in Olfactory Disorders

During this period, 1000 participants > 18 y/o will be recruited and divided into four distinct groups: normosmics (550), hyposmics (150), tinnitus patients with hyposmia (150), and tinnitus patients without hyposmia (150). The primary outcome of the Neuroscent project is to assess the ability of EEG analysis to distinguish between normosmic and hyposmic participants during olfactory stimulation using the Sniffin' Sticks test.

Alpha Power: Measured in microvolts squared (µV²), representing the magnitude of Alpha wave activity during olfactory stimulation. Theta Power: Measured in microvolts squared (µV²), representing the magnitude of Theta wave activity during olfactory stimulation. We will make use of the Fourier transform as well, which is inherently dimensionless. The data acquisition setup involves placing blindfolded participants in an isolated room and administering three randomized olfactory stimuli (pure olfactory, trigeminal, and olfactory-gustatory) while collecting EEG data. Sessions start and end with a 30-second resting state.

Olfactory Threshold Scores (as per routinary diagnosis of olfactory performances): Derived from the Sniffin' Sticks test, these scores quantify participants' ability to detect odors, classifying them as normosmic or hyposmic. All the participants will undergo a complete olfactory assessment through the Sniffin' Sticks test. These are represented as numerical scores. The maximum score obtainable in the full Sniffin' Sticks test, which includes the three sub-tests (Threshold, Discrimination, Identification - TDI), is 48. This maximum score is achieved by summing the highest possible score from each of the three components: 16 for olfactory threshold (Threshold), 16 for olfactory discrimination (Discrimination), and 16 for olfactory identification (Identification).

Random Forest Model Classification Accuracy: The accuracy of the classification model used to distinguish between normosmic and hyposmic participants based on EEG data, usually expressed as a percentage or a value between 0 and 1.

t-SNE Features: Although these don't have a specific unit, they represent the reduced dimensional data points used as features in the Random Forest model. These are typically dimensionless and used in the context of visualizing and improving model performance.

Pure Tone Audiometry (PTA), as per routinary diagnosis of Tinnitus: Measures hearing thresholds across frequencies, reported in decibels Hearing Level (dB HL) to determine the degree of hearing loss. The results are plotted on an audiogram, which shows the person's hearing threshold at different frequencies; a normal audiogram threshold is 0-20 dB HL between 250 and 8k Hz.

Tinnitus Handicap Inventory (THI): Scores the impact of tinnitus on a scale from 0 to 100, where a higher score indicates a worse impact of tinnitus.

EEG Data: Analyzed during exposure to olfactory stimuli (as per Primary Outcome), with features reduced via t-SNE and classified using a Random Forest algorithm.

Random Forest Model Classification Accuracy: Measures the accuracy of the model used to classify participants across the four groups. This is typically expressed as a percentage or a value between 0 and 1. Fleiss' Kappa Statistic: Used to evaluate the agreement of classification, reported as a kappa coefficient.

Power Analysis: Ensures the sample size is adequate to detect significant effects. This is typically expressed in terms of statistical power (e.g., 0.8 or 80%). Post-hoc Analyses: Conducted if significant differences are found, usually involving p-values or effect sizes.

Auditory Brainstem Response (ABR) as per routinary diagnosis of Tinnitus: Assesses the auditory nerve and brainstem's response to sound stimuli by measuring electrical activity in the brain. Results are typically presented as waveforms, with particular focus on latency and amplitude of specific waves, measured in milliseconds (ms) and microvolts (µV) respectively, to evaluate neural pathways involved in hearing.

Tympanometry, as per routinary diagnosis of Tinnitus: Evaluates middle ear function by measuring tympanic membrane mobility, with results reported in decapascals (daPa) and milliliters (mL). The pressure reading in tympanometry reflects the air pressure in the middle ear cavity. A normal tympanogram shows peak compliance when the pressure in the ear canal matches the pressure in the middle ear, indicating normal middle ear function. The two units of measurement, pressure and compliance, are plotted on a two-dimensional graph. Pressure (daPa) is plotted on the x-axis and compliance, measured in mL, is plotted on the y-axis in the two-dimensional graph

The final outcome will involve comparing olfactory perception across four participant groups (normosmic, normosmic with tinnitus, hyposmic, and hyposmic with tinnitus) using EEG features derived from Fourier analysis, reduced via t-SNE, and classified with a Random Forest algorithm. The model's accuracy in predicting odor categories (based on EEG data) will be evaluated using metrics such as accuracy, precision, recall, specificity, and F1 score. Fleiss' kappa statistic will assess agreement across groups, comparing the algorithm's predictions with the Sniffin' Sticks test results. Statistical power analysis ensures adequate sample sizes for reliable comparisons, with post-hoc analyses conducted if significant differences are detected.