Pure-tone audiometry (PTA) is a basic form of hearing test which assess hearing thresholds across various frequencies of sound. It is the gold standard for hearing tests and commonly used across the world. PTA’s provide meaningful insights into an individual’s hearing profile and can be used for diagnostics of possible types and sources of hearing loss. From this test, an audiogram can be generated which visualises an individual’s hearing profile (Figure 1.). Audiogram curves are a crucial form of hearing health data.

Fig 1. Example audiogram curve from The National Health and Nutrition Examination Survey.

The National Health and Nutrition Examination Survey (NHANES) is the largest openly available dataset for pure-tone audiometry (PTA) data. Here, audiogram curves from NHANES 2011–2012, 2015–2016, and 2017–2020 were used to create a variational model of audiogram data. This model was sampled from, providing new, synthetic observations of audiogram data (Figure 2.). These data are statistically equivalent to the original data while also not being related to an actual individuals. A similar approach can be taken with University College London Hospital’s (UCLH) data to provide synthetic versions of UCLH’s audiogram data which is had risk of personal, private, or sensitive information leakage. This procedure generates data that cannot be used to identify an individual.

Fig 2. Example synthetic audiogram curve generate by model.

This audiogram curve is “audiologically plausible” yet, is completely synthetic and does not pertain to any individual.