Single-neuron representations of odours in the human brain.

Olfaction is a fundamental sensory modality that guides animal and human behaviour1,2. However, the underlying neural processes of human olfaction are still poorly understood at the fundamental-that is, the single-neuron-level. Here we report recordings of single-neuron activity in the piriform cortex and medial temporal lobe in awake humans performing an odour rating and identification task. We identified odour-modulated neurons within the piriform cortex, amygdala, entorhinal cortex and hippocampus. In each of these regions, neuronal firing accurately encodes odour identity. Notably, repeated odour presentations reduce response firing rates, demonstrating central repetition suppression and habituation. Different medial temporal lobe regions have distinct roles in odour processing, with amygdala neurons encoding subjective odour valence, and hippocampal neurons predicting behavioural odour identification performance. Whereas piriform neurons preferably encode chemical odour identity, hippocampal activity reflects subjective odour perception. Critically, we identify that piriform cortex neurons reliably encode odour-related images, supporting a multimodal role of the human piriform cortex. We also observe marked cross-modal coding of both odours and images, especially in the amygdala and piriform cortex. Moreover, we identify neurons that respond to semantically coherent odour and image information, demonstrating conceptual coding schemes in olfaction. Our results bridge the long-standing gap between animal models and non-invasive human studies and advance our understanding of odour processing in the human brain by identifying neuronal odour-coding principles, regional functional differences and cross-modal integration.

Covid-19 affects taste independent of taste-smell confusions: results from a combined chemosensory home test and online survey from a large global cohort.

People often confuse smell loss with taste loss, so it is unclear how much gustatory function is reduced in patients self-reporting taste loss. Our pre-registered cross-sectional study design included an online survey in 12 languages with instructions for self-administering chemosensory tests with 10 household items. Between June 2020 and March 2021, 10,953 individuals participated. Of these, 5,225 self-reported a respiratory illness and were grouped based on their reported COVID test results: COVID-positive (COVID+, N = 3,356), COVID-negative (COVID-, N = 602), and COVID unknown for those waiting for a test result (COVID?, N = 1,267). The participants who reported no respiratory illness were grouped by symptoms: sudden smell/taste changes (STC, N = 4,445), other symptoms excluding smell or taste changes (OthS, N = 832), and no symptoms (NoS, N = 416). Taste, smell, and oral irritation intensities and self-assessed abilities were rated on visual analog scales. Compared to the NoS group, COVID+ was associated with a 21% reduction in taste (95% confidence interval (CI): 15-28%), 47% in smell (95% CI: 37-56%), and 17% in oral irritation (95% CI: 10-25%) intensity. There were medium to strong correlations between perceived intensities and self-reported abilities (r = 0.84 for smell, r = 0.68 for taste, and r = 0.37 for oral irritation). Our study demonstrates that COVID-19-positive individuals report taste dysfunction when self-tested with stimuli that have little to none olfactory components. Assessing the smell and taste intensity of household items is a promising, cost-effective screening tool that complements self-reports and may help to disentangle taste loss from smell loss. However, it does not replace standardized validated psychophysical tests.

Recent Smell Loss Is the Best Predictor of COVID-19 Among Individuals With Recent Respiratory Symptoms.

In a preregistered, cross-sectional study, we investigated whether olfactory loss is a reliable predictor of COVID-19 using a crowdsourced questionnaire in 23 languages to assess symptoms in individuals self-reporting recent respiratory illness. We quantified changes in chemosensory abilities during the course of the respiratory illness using 0-100 visual analog scales (VAS) for participants reporting a positive (C19+; n = 4148) or negative (C19-; n = 546) COVID-19 laboratory test outcome. Logistic regression models identified univariate and multivariate predictors of COVID-19 status and post-COVID-19 olfactory recovery. Both C19+ and C19- groups exhibited smell loss, but it was significantly larger in C19+ participants (mean ± SD, C19+: -82.5 ± 27.2 points; C19-: -59.8 ± 37.7). Smell loss during illness was the best predictor of COVID-19 in both univariate and multivariate models (ROC AUC = 0.72). Additional variables provide negligible model improvement. VAS ratings of smell loss were more predictive than binary chemosensory yes/no-questions or other cardinal symptoms (e.g., fever). Olfactory recovery within 40 days of respiratory symptom onset was reported for ~50% of participants and was best predicted by time since respiratory symptom onset. We find that quantified smell loss is the best predictor of COVID-19 amongst those with symptoms of respiratory illness. To aid clinicians and contact tracers in identifying individuals with a high likelihood of having COVID-19, we propose a novel 0-10 scale to screen for recent olfactory loss, the ODoR-19. We find that numeric ratings ≤2 indicate high odds of symptomatic COVID-19 (4 < OR < 10). Once independently validated, this tool could be deployed when viral lab tests are impractical or unavailable.

More Than Smell-COVID-19 Is Associated With Severe Impairment of Smell, Taste, and Chemesthesis.

Recent anecdotal and scientific reports have provided evidence of a link between COVID-19 and chemosensory impairments, such as anosmia. However, these reports have downplayed or failed to distinguish potential effects on taste, ignored chemesthesis, and generally lacked quantitative measurements. Here, we report the development, implementation, and initial results of a multilingual, international questionnaire to assess self-reported quantity and quality of perception in 3 distinct chemosensory modalities (smell, taste, and chemesthesis) before and during COVID-19. In the first 11 days after questionnaire launch, 4039 participants (2913 women, 1118 men, and 8 others, aged 19-79) reported a COVID-19 diagnosis either via laboratory tests or clinical assessment. Importantly, smell, taste, and chemesthetic function were each significantly reduced compared to their status before the disease. Difference scores (maximum possible change ±100) revealed a mean reduction of smell (-79.7 ± 28.7, mean ± standard deviation), taste (-69.0 ± 32.6), and chemesthetic (-37.3 ± 36.2) function during COVID-19. Qualitative changes in olfactory ability (parosmia and phantosmia) were relatively rare and correlated with smell loss. Importantly, perceived nasal obstruction did not account for smell loss. Furthermore, chemosensory impairments were similar between participants in the laboratory test and clinical assessment groups. These results show that COVID-19-associated chemosensory impairment is not limited to smell but also affects taste and chemesthesis. The multimodal impact of COVID-19 and the lack of perceived nasal obstruction suggest that severe acute respiratory syndrome coronavirus strain 2 (SARS-CoV-2) infection may disrupt sensory-neural mechanisms.

Prefrontal Control Over Occipital Responses to Crossmodal Overlap Varies Across the Congruency Spectrum.

While matched crossmodal information is known to facilitate object recognition, it is unclear how our perceptual systems encode the more gradual congruency variations that occur in our natural environment. Combining visual objects with odor mixtures to create a gradual increase in semantic object overlap, we demonstrate high behavioral acuity to linear variations of olfactory-visual overlap in a healthy adult population. This effect was paralleled by a linear increase in cortical activation at the intersection of occipital fusiform and lingual gyri, indicating linear encoding of crossmodal semantic overlap in visual object recognition networks. Effective connectivity analyses revealed that this integration of olfactory and visual information was achieved by direct information exchange between olfactory and visual areas. In addition, a parallel pathway through the superior frontal gyrus was increasingly recruited towards the most ambiguous stimuli. These findings demonstrate that cortical structures involved in object formation are inherently crossmodal and encode sensory overlap in a linear manner. The results further demonstrate that prefrontal control of these processes is likely required for ambiguous stimulus combinations, a fact of high ecological relevance that may be inappropriately captured by common task designs juxtaposing congruency and incongruency.

A bittersweet symphony: Evidence for taste-sound correspondences without effects on taste quality-specific perception.

Music has been associated with taste and shown to influence the dining experience. We asked whether sound that is associated with taste affects taste perception of food. In two studies (study 1: N = 20, 13 women; study 2: N = 20, 17 women), participants evaluated the taste of cinder toffee while listening to either of two soundscapes associated with sweet and bitter taste, respectively, or no sound. In study 1, participants rated the taste on a visual-analog scale (VAS) anchored with "bitter" and "sweet", aiming to replicate a previous study (Crisinel et al., ). In contrast, four separate scales were used in study 2 to report the extent of bitter, sweet, sour, and salty taste to test whether taste qualities were influenced by sound differentially. Additionally, taste intensity and pleasantness were rated in both studies. Taste intensity was increased in the presence of a sound, while pleasantness was not affected. In study 1, sound shifted bitter-sweet ratings in the direction of the congruent sound, i.e. samples tasted sweeter with "sweet" sound and more bitter with "bitter" sound, replicating Crisinel et al.'s () results. However, this effect was abolished when a "no-sound" control was included in the statistical model. Taste ratings in study 2 showed no effect of sound on any specific taste quality, suggesting that the influence of sound on taste in study 1 reflects an artifact of the scale rather than an actual shift in perception. Together, the data provide evidence for taste-sound correspondences without effects on taste-quality specific perception.

Recognizing Taste: Coding Patterns Along the Neural Axis in Mammals.

The gustatory system encodes information about chemical identity, nutritional value, and concentration of sensory stimuli before transmitting the signal from taste buds to central neurons that process and transform the signal. Deciphering the coding logic for taste quality requires examining responses at each level along the neural axis-from peripheral sensory organs to gustatory cortex. From the earliest single-fiber recordings, it was clear that some afferent neurons respond uniquely and others to stimuli of multiple qualities. There is frequently a "best stimulus" for a given neuron, leading to the suggestion that taste exhibits "labeled line coding." In the extreme, a strict "labeled line" requires neurons and pathways dedicated to single qualities (e.g., sweet, bitter, etc.). At the other end of the spectrum, "across-fiber," "combinatorial," or "ensemble" coding requires minimal specific information to be imparted by a single neuron. Instead, taste quality information is encoded by simultaneous activity in ensembles of afferent fibers. Further, "temporal coding" models have proposed that certain features of taste quality may be embedded in the cadence of impulse activity. Taste receptor proteins are often expressed in nonoverlapping sets of cells in taste buds apparently supporting "labeled lines." Yet, taste buds include both narrowly and broadly tuned cells. As gustatory signals proceed to the hindbrain and on to higher centers, coding becomes more distributed and temporal patterns of activity become important. Here, we present the conundrum of taste coding in the light of current electrophysiological and imaging techniques at several levels of the gustatory processing pathway.

A new gustometer: Template for the construction of a portable and modular stimulator for taste and lingual touch.

Taste research has been hampered by technical difficulties, mostly because liquid taste stimuli are difficult to control in terms of timing and application area. Exact stimulus control requires a gustometer, but the existing devices are either not well-documented or rather inflexible. We designed a gustometer based on a computer-controlled, modular pump system, which can be extended via additional hardware modules-for example, for heating of the stimuli or sending and receiving triggers. All components are available for purchase "off the shelf." The pumps deliver liquids through plastic tubing and can be connected to commercially available or custom-made mouthpieces. We determined the temporal precision of the device. Onset delays showed minuscule variation within pumps (SD < 3 ms) and small differences between pumps (< 4.5 ms). The rise time was less than 2 ms (SD < 2 ms), and the dosage volume bias was only 2%. To test whether hemitongues could be stimulated independently, we conducted a behavioral experiment. A total of 18 participants received tasteless stimuli to the left, right, or both sides of the tongue. The side of stimulation was correctly identified on 91% of trials, indicating that the setup is suitable for lateralized stimulation. Electroencephalographic responses to water and salty stimuli were recorded from two participants; the stimulation successfully evoked event-related responses, demonstrating the suitability of the device for use in electrophysiological investigations. We provide a Python-based open-source software package and a Web interface to easily operate the system. We thereby hope to facilitate access to state-of-the-art taste research methods and to increase reproducibility across laboratories.

Delta activity encodes taste information in the human brain.

The categorization of food via sensing nutrients or toxins is crucial to the survival of any organism. On ingestion, rapid responses within the gustatory system are required to identify the oral stimulus to guide immediate behavior (swallowing or expulsion). The way in which the human brain accomplishes this task has so far remained unclear. Using multivariate analysis of 64-channel scalp EEG recordings obtained from 16 volunteers during tasting salty, sweet, sour, or bitter solutions, we found that activity in the delta-frequency range (1-4 Hz; delta power and phase) has information about taste identity in the human brain, with discriminable response patterns at the single-trial level within 130 ms of tasting. Importantly, the latencies of these response patterns predicted the point in time at which participants indicated detection of a taste by pressing a button. Furthermore, taste pattern discrimination was independent of motor-related activation and encoded taste identity rather than other taste features such as intensity and valence. On comparison with our previous findings from a delayed taste-discrimination task (Crouzet et al., 2015), taste-specific neural representations emerged earlier during this speeded taste-detection task, suggesting a goal-dependent flexibility in gustatory response coding. Together, these findings provide the first evidence of a role of delta activity in taste-information coding in humans. Crucially, these neuronal response patterns can be linked to the speed of simple gustatory perceptual decisions - a vital performance index of nutrient sensing.

Visual-Olfactory Interactions: Bimodal Facilitation and Impact on the Subjective Experience.

Odors are inherently ambiguous and therefore susceptible to redundant sensory as well as context information. The identification of an odor object relies largely on visual input. Thus far, it is unclear whether visual and olfactory stimuli are indeed integrated at an early perceptual stage and which role the congruence between the visual and olfactory inputs plays. Previous studies on visual-olfactory interaction used either congruent or incongruent information, leaving it open whether nuances of visual-olfactory congruence shape perception differently. We aimed to answer 1) whether visual-olfactory information is integrated at early stages of processing, 2) whether visual-olfactory congruence is a gradual or dichotomous phenomenon, and 3) whether visual information influences bimodal stimulus evaluation and odor identity. We found a bimodal response time speedup that is consistent with parallel processing according to race models. Subjectively, pleasantness of bimodal stimuli increased with increasing congruence, and orange images biased odor composition toward orange. Visual-olfactory congruence was perceived in gradual and distinct categories, consistent with the notion that congruence is a gradual phenomenon. Together, the data provide evidence for bimodal facilitation consistent with parallel processing of the visual and olfactory stimuli, and that visual-olfactory interactions influence various levels of the subjective experience.

The NutriAct Family Study: a web-based prospective study on the epidemiological, psychological and sociological basis of food choice.

BACKGROUND: Most studies on food choice have been focussing on the individual level but familial aspects may also play an important role. This paper reports of a novel study that will focus on the familial aspects of the formation of food choice among men and women aged 50-70 years by recruiting spouses and siblings (NutriAct Family Study; NFS). METHODS: Data is collected prospectively via repeatedly applied web-based questionnaires over the next years. The recruitment for the NFS started in October 2016. Participants are recruited based on an index person who is actively participating in the European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam study. This index person was asked to invite the spouse, a sibling or an in-law. If a set of family members agreed to participate, access to individualized web-based questionnaires assessing dietary intake, other health related lifestyle habits, eating behaviour, food responsiveness, personality, self-regulation, socio-economic status and socio-cultural values was provided. In the first phase of the NSF, recruitment rates were monitored in detail and participants' comments were analysed in order to improve the feasibility of procedures and instruments. DISCUSSION: Until August 4th 2017, 4783 EPIC-Participants were contacted by mail of which 446 persons recruited 2 to 5 family members (including themselves) resulting in 1032 participants, of whom 82% had started answering or already completed the questionnaires. Of the 4337 remaining EPIC-participants who had been contacted, 1040 (24%) did not respond at all, and 3297 (76%) responded but declined, in 51% of the cases because of the request to recruit at least 2 family members in the respective age range. The developed recruitment procedures and web-based methods of data collection are capable to generate the required study population including the data on individual and inter-personal determinants which will be linkable to food choice. The information on familial links among the study participants will show the role of familial traits in midlife for the adoption of food choices supporting healthy aging.

Changes in Gustatory Function and Taste Preference Following Weight Loss.

OBJECTIVE: To investigate taste changes of obese children during an inpatient weight reduction treatment in comparison with normal weight children. STUDY DESIGN: Obese (n = 60) and normal weight (n = 27) children aged 9-17 years were assessed for gustatory functions using taste strips (taste identification test for the taste qualities sour, salty, sweet, and bitter), taste preferences, and experienced taste sensitivity. Obese children were examined upon admission (T1) and before discharge (T2). Normal weight children served as the control group. RESULTS: Irrespective of taste quality, obese children exhibited a lower ability to identify taste (total taste score) than normal weight children (P < .01); this overall score remained stable during inpatient treatment in obese children. Group and treatment effects were seen when evaluating individual taste qualities. In comparison with normal weight children, obese children exhibited poorer sour taste identification performance (P < .01). Obese children showed improvement in sour taste identification (P < .001) and deterioration in sweet taste identification (P < .001) following treatment. Subjective reports revealed a lower preference for sour taste in obese children compared with normal weight children (P < .05). The sweet and bitter taste ability at T1 predicted the body mass index z score at T2 (R2 = .23, P < .01). CONCLUSIONS: We identified differences in the ability to discriminate tastes and in subjective taste perception between groups. Our findings of increased sour and reduced sweet taste discrimination after the intervention in obese children are indicative of an exposure-related effect on taste performance, possibly mediated by increased acid and reduced sugar consumption during the intervention. Because the sweet and bitter taste ability at T1 predicted weight loss, addressing gustatory function could be relevant in individualized obesity treatment approaches. TRIAL REGISTRATION: Germanctr.de: DRKS00005122.

Superadditive and Subadditive Neural Processing of Dynamic Auditory-Visual Objects in the Presence of Congruent Odors.

Our sensory experiences comprise a variety of different inputs at any given time. Some of these experiences are unmistakable, others are ambiguous and profit from additional sensory information. Here, we explored whether the presence of a congruent odor influences the neural processing and sensory interaction of audio-visual objects using degraded videos (V) and sounds (A) of dynamic objects in unimodal and bimodal (AV) combinations without or with a congruent odor (VO, AO, AVO). Analyses of EEG data revealed superadditive and subadditive interaction effects. The topography and timing of these effects suggest evaluative rather than sensory processes as the underlying cause. Together, the results suggest that the mere presence of an odor affects the processing of A, V, and AV objects differently while multisensory interactions of AV and AVO objects have common neuronal mechanisms pointing to a robust, modality-independent network for the processing of redundant sensory information.

Higher sensitivity to sweet and salty taste in obese compared to lean individuals.

Although putatively taste has been associated with obesity as one of the factors governing food intake, previous studies have failed to find a consistent link between taste perception and Body Mass Index (BMI). A comprehensive comparison of both thresholds and hedonics for four basic taste modalities (sweet, salty, sour, and bitter) has only been carried out with a very small sample size in adults. In the present exploratory study, we compared 23 obese (OB; BMI > 30), and 31 lean (LN; BMI < 25) individuals on three dimensions of taste perception - recognition thresholds, intensity, and pleasantness - using different concentrations of sucrose (sweet), sodium chloride (NaCl; salty), citric acid (sour), and quinine hydrochloride (bitter) dissolved in water. Recognition thresholds were estimated with an adaptive Bayesian staircase procedure (QUEST). Intensity and pleasantness ratings were acquired using visual analogue scales (VAS). It was found that OB had lower thresholds than LN for sucrose and NaCl, indicating a higher sensitivity to sweet and salty tastes. This effect was also reflected in ratings of intensity, which were significantly higher in the OB group for the lower concentrations of sweet, salty, and sour. Calculation of Bayes factors further corroborated the differences observed with null-hypothesis significance testing (NHST). Overall, the results suggest that OB are more sensitive to sweet and salty, and perceive sweet, salty, and sour more intensely than LN.

Perceived Odor-Taste Congruence Influences Intensity and Pleasantness Differently.

The role of congruence in cross-modal interactions has received little attention. In most experiments involving cross-modal pairs, congruence is conceived of as a binary process according to which cross-modal pairs are categorized as perceptually and/or semantically matching or mismatching. The present study investigated whether odor-taste congruence can be perceived gradually and whether congruence impacts other facets of subjective experience, that is, intensity, pleasantness, and familiarity. To address these questions, we presented food odorants (chicken, orange, and 3 mixtures of the 2) and tastants (savory-salty and sour-sweet) in pairs varying in congruence. Participants were to report the perceived congruence of the pairs along with intensity, pleasantness, and familiarity. We found that participants could perceive distinct congruence levels, thereby favoring a multilevel account of congruence perception. In addition, familiarity and pleasantness followed the same pattern as the congruence while intensity was highest for the most congruent and the most incongruent pairs whereas intensities of the intermediary-congruent pairs were reduced. Principal component analysis revealed that pleasantness and familiarity form one dimension of the phenomenological experience of odor-taste pairs that was orthogonal to intensity. The results bear implications for the understanding the behavioral underpinnings of perseverance of habitual food choices.

Superadditive opercular activation to food flavor is mediated by enhanced temporal and limbic coupling.

Food perception is characterized by a transition from initially separate sensations of the olfactory and gustatory properties of the object toward their combined sensory experience during consumption. The holistic flavor experience, which occurs as the smell and taste merge, extends beyond the mere addition of the two chemosensory modalities, being usually perceived as more object-like, intense and rewarding. To explore the cortical mechanisms which give rise to olfactory-gustatory binding during natural food consumption, brain activation during consumption of a pleasant familiar beverage was contrasted with presentation of its taste and orthonasal smell alone. Convergent activation to all presentation modes was observed in executive and chemosensory association areas. Flavor, but not orthonasal smell or taste alone, stimulated the frontal operculum, supporting previous accounts of its central role in the formation of the flavor percept. A functional dissociation was observed in the insula: the anterior portion was characterized by sensory convergence, while mid-dorsal sections activated exclusively to the combined flavor stimulus. psycho-physiological interaction analyses demonstrated increased neural coupling between the frontal operculum and the anterior insula during flavor presentation. Connectivity was also increased with the lateral entorhinal cortex, a relay to memory networks and central node for contextual modulation of olfactory processing. These findings suggest a central role of the insular cortex in the transition from mere detection of chemosensory convergence to a superadditive flavor representation. The increased connections between the frontal operculum and medial temporal memory structures during combined olfactory-gustatory stimulation point to a potential mechanism underlying the acquisition and modification of flavor preferences.

Verbal labels selectively bias brain responses to high-energy foods.

The influence of external factors on food preferences and choices is poorly understood. Knowing which and how food-external cues impact the sensory processing and cognitive valuation of food would provide a strong benefit toward a more integrative understanding of food intake behavior and potential means of interfering with deviant eating patterns to avoid detrimental health consequences for individuals in the long run. We investigated whether written labels with positive and negative (as opposed to 'neutral') valence differentially modulate the spatio-temporal brain dynamics in response to the subsequent viewing of high- and low-energetic food images. Electrical neuroimaging analyses were applied to visual evoked potentials (VEPs) from 20 normal-weight participants. VEPs and source estimations in response to high- and low- energy foods were differentially affected by the valence of preceding word labels over the ~260-300 ms post-stimulus period. These effects were only observed when high-energy foods were preceded by labels with positive valence. Neural sources in occipital as well as posterior, frontal, insular and cingulate regions were down-regulated. These findings favor cognitive-affective influences especially on the visual responses to high-energetic food cues, potentially indicating decreases in cognitive control and goal-adaptive behavior. Inverse correlations between insular activity and effectiveness in food classification further indicate that this down-regulation directly impacts food-related behavior.

Nutrition claims influence expectations about food attributes, attenuate activity in reward-associated brain regions during tasting, but do not impact pleasantness.

INTRODUCTION: Nutrition claims are one of the most common tools used to improve food decisions. Previous research has shown that nutrition claims impact expectations; however, their effects on perceived pleasantness, valuation, and their neural correlates are not well understood. These claims may have both intended and unintended effects on food perception and valuation, which may compromise their effect on food decisions. METHODS: We investigated the effects of nutrition claims on expectations, perceptions, and valuation of milk-mix drinks in a behavioral (n = 110) and an fMRI (n = 39) study. In the behavioral study, we assessed the effects of a "fat-reduced" and a "protein-rich" nutrition claim on expected and perceived food attributes of otherwise equal food products. In the fMRI study, we investigated the effect of a "protein-rich" claim on taste pleasantness perception and valuation, and on their neural correlates during tasting and swallowing. RESULTS: We found that both nutrition claims increased expected and perceived healthiness and decreased expected but not perceived taste pleasantness. The "protein-rich" claim increased expected but not perceived satiating quality ratings, while the "fat-reduced" claim decreased both expected and perceived satiating quality ratings. In the absence vs. presence of the "protein-rich" claim, we observed an increased activity in a cluster extending to the left nucleus accumbens during tasting and an increased functional connectivity between this cluster and a cluster in right middle frontal gyrus during swallowing. CONCLUSION: Altogether, we found that nutrition claims impacted expectations and attenuated reward-related responses during tasting but did not negatively affect perceived pleasantness. Our findings support highlighting the presence of nutrients with positive associations and exposure to foods with nutrition claims to increase their acceptance. Our study offers insights that may be valuable in designing and optimizing the use of nutrition claims.

Text-based predictions of COVID-19 diagnosis from self-reported chemosensory descriptions.

BACKGROUND: There is a prevailing view that humans' capacity to use language to characterize sensations like odors or tastes is poor, providing an unreliable source of information. METHODS: Here, we developed a machine learning method based on Natural Language Processing (NLP) using Large Language Models (LLM) to predict COVID-19 diagnosis solely based on text descriptions of acute changes in chemosensation, i.e., smell, taste and chemesthesis, caused by the disease. The dataset of more than 1500 subjects was obtained from survey responses early in the COVID-19 pandemic, in Spring 2020. RESULTS: When predicting COVID-19 diagnosis, our NLP model performs comparably (AUC ROC ~ 0.65) to models based on self-reported changes in function collected via quantitative rating scales. Further, our NLP model could attribute importance of words when performing the prediction; sentiment and descriptive words such as "smell", "taste", "sense", had strong contributions to the predictions. In addition, adjectives describing specific tastes or smells such as "salty", "sweet", "spicy", and "sour" also contributed considerably to predictions. CONCLUSIONS: Our results show that the description of perceptual symptoms caused by a viral infection can be used to fine-tune an LLM model to correctly predict and interpret the diagnostic status of a subject. In the future, similar models may have utility for patient verbatims from online health portals or electronic health records.

Early in the COVID-19 pandemic, people who were infected with SARS-CoV-2 reported changes in smell and taste. To better study these symptoms of SARS-CoV-2 infections and potentially use them to identify infected patients, a survey was undertaken in various countries asking people about their COVID-19 symptoms. One part of the questionnaire asked people to describe the changes in smell and taste they were experiencing. We developed a computational program that could use these responses to correctly distinguish people that had tested positive for SARS-CoV-2 infection from people without SARS-CoV-2 infection. This approach could allow rapid identification of people infected with SARS-CoV-2 from descriptions of their sensory symptoms and be adapted to identify people infected with other viruses in the future.