This odor detection threshold test pairs a novel adaptive algorithm with a card-based tool to rapidly determine an individual’s odor detection threshold. Potential uses include: (1) assessing changes in an individual’s ability to smell over time or under different conditions; (2) determining whether an individual’s sense of smell meets criteria for study participation; and (3) diagnosing a quantitative olfactory disorder. Olfactory dysfunctions entail the loss or distortion of the sense of smell, and are prevalent, underdiagnosed medical conditions. They have significant consequences for health, diet, safety, and quality of life, including impaired ability to detect dangers, like fires and soiled foods, altered diets, and disconnection from the environment. Estimates suggest up to 1 in 4 people have an olfactory disorder, commonly caused by head trauma, sinonasal disease, and upper respiratory infections.
In clinical settings, doctors commonly use psychophysical testing to determine if an individual has quantitative olfactory disorders. These disorders include anosmia, a complete or nearly complete loss of smell, and hyposmia, a substantially reduced ability to detect odors. Quantitative tests measure one or more smell function parameters such as odor identification, discrimination, or detection threshold. The two most common tests in clinical and research settings are the University of Pennsylvania Smell Identification Test (UPSIT), featuring 40 odor identification questions, and Sniffin’ Sticks, which measures odor identification, discrimination, and detection threshold. However, these widely and commonly used tests have several limitations. For example, odor identification tests require significant cognitive demands, such as recognizing the stimulus from prior experience and communicating the correct name, and may include odors that are unfamiliar to younger patients or individuals from different cultures. The Sniffin’ Sticks test requires a trained test administrator, a person who is often not available outside of a research lab or specialized clinic. Both tests take too long (20-40 min) for a typical visit to the doctor’s office. These limitations lead to limited testing of olfactory dysfunction in routine clinical practice, resulting in a rapidly growing demand for diagnostic tools.
Researchers at the University of Florida have developed the Adaptive Olfactory Measure of Threshold (ArOMa-T) test, an odor detection test that measures olfactory function using an adaptive algorithm. The ArOMa-T test, which can be self-administered, pairs a card-based odor-delivery device with an app that contains the algorithm and directs the user through the test. The ArOMa-T is rapid, robust, inexpensive, straightforward to understand, and easily performed almost anywhere.
ArOMa-T uses an adaptive algorithm and pairs with a card-based tool to test for olfactory dysfunction simply, rapidly, and inexpensively
The Adaptive Olfactory Measure of Threshold (ArOMa-T) test pairs a mobile app with a bi-fold card to evaluate an individual’s odor detection threshold. The card contains graphics and text on the outside face with user instructions and 17 peel-and-burst labels containing an odorant on the interior. The peel-and-burst delivery systems enable the release of the same odorant concentration in every test, leading to consistent results. The labels of the initial version of the ArOMa-T card contain different concentrations of the floral odorant phenyl ethyl alcohol (PEA), widely used in smell testing, although other odorants can be used. Of labels 1-16, fourteen contain different PEA concentrations, while two have no added odorant.
The ArOMa-T test also employs an adaptive threshold estimation algorithm utilizing a Bayesian model to determine which label the user should peel and sniff next. The accompanying app guides the participants through the tasks, asking them to sniff Label 0 first as a reference smell of the card and label. Then, the algorithm directs the participant to sniff label no. 1 and record whether they can smell it. Depending on the response, the algorithm guides the participant to sniff a specific label most likely to reduce uncertainty in the running estimate of the detection threshold parameter, repeating this process for the rest of the trials. To decrease unfaithful responses, there are duplicates of odor concentrations and blanks. As the test progresses, the algorithm fits a psychometric curve estimating the probability of a “Yes” response at all concentrations, and the uncertainty of this curve decreases with the addition of trials. The final detection threshold estimates stem from eight odorant samples, including the blank labels.