Leveraging Machine Learning and Self-Administered Tests to Predict Covid-19: an Olfactory and Gustatory Dysfunction Assessment Through Crowdsourced Data in India

Olfactory and gustatory dysfunction are common manifestations associated with COVID-19 infection, typically identified through medical evaluations or self-reports. However, self-reports are often complicated by the common confusion between these senses. In this study, we explored whether practical smell and taste tests using household items could help assess these sensory losses. Participants completed an online survey, performing smell and taste tests with everyday items. We also collected demographic data, health. These authors have equally contributed to this work. information (including COVID-19 diagnosis), and current symptoms. Using various combinations of these factors, we built multiple random forest models to predict COVID-19 diagnosis. Incorporating data from the smell and taste self-tests enhanced the predictive accuracy of the models. Certain household items like clove for smell and lemon juice for taste demonstrated statistically significant capability in distinguishing infected ones from non-infected ones. This study demonstrates that simple smell and taste tests using household items can help predict COVID-19 infection, offering a potential home-based method to identify cases and limit transmission.