Giving a Voice to Patients With Smell Disorders Associated With COVID-19: Cross-Sectional Longitudinal Analysis Using Natural Language Processing of Self-Reports
Bibliographic record
Abstract
BACKGROUND: Smell disorders are commonly reported with COVID-19 infection. The smell-related issues associated with COVID-19 may be prolonged, even after the respiratory symptoms are resolved. These smell dysfunctions can range from anosmia (complete loss of smell) or hyposmia (reduced sense of smell) to parosmia (smells perceived differently) or phantosmia (smells perceived without an odor source being present). Similar to the difficulty that people experience when talking about their smell experiences, patients find it difficult to express or label the symptoms they experience, thereby complicating diagnosis. The complexity of these symptoms can be an additional burden for patients and health care providers and thus needs further investigation. OBJECTIVE: This study aims to explore the smell disorder concerns of patients and to provide an overview for each specific smell disorder by using the longitudinal survey conducted in 2020 by the Global Consortium for Chemosensory Research, an international research group that has been created ad hoc for studying chemosensory dysfunctions. We aimed to extend the existing knowledge on smell disorders related to COVID-19 by analyzing a large data set of self-reported descriptive comments by using methods from natural language processing. METHODS: We included self-reported data on the description of changes in smell provided by 1560 participants at 2 timepoints (second survey completed between 23 and 291 days). Text data from participants who still had smell disorders at the second timepoint (long-haulers) were compared with the text data of those who did not (non-long-haulers). Specifically, 3 aims were pursued in this study. The first aim was to classify smell disorders based on the participants' self-reports. The second aim was to classify the sentiment of each self-report by using a machine learning approach, and the third aim was to find particular food and nonfood keywords that were more salient among long-haulers than those among non-long-haulers. RESULTS: We found that parosmia (odds ratio [OR] 1.78, 95% CI 1.35-2.37; P<.001) as well as hyposmia (OR 1.74, 95% CI 1.34-2.26; P<.001) were more frequently reported in long-haulers than in non-long-haulers. Furthermore, a significant relationship was found between long-hauler status and sentiment of self-report (P<.001). Finally, we found specific keywords that were more typical for long-haulers than those for non-long-haulers, for example, fire, gas, wine, and vinegar. CONCLUSIONS: Our work shows consistent findings with those of previous studies, which indicate that self-reports, which can easily be extracted online, may offer valuable information to health care and understanding of smell disorders. At the same time, our study on self-reports provides new insights for future studies investigating smell disorders.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.005 | 0.012 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".