Multidimensional Active Testing

Videos to Illustrate Multidimensional Active Testing (MAT)

Example of traditional, inefficient method of independent threshold testing for each test condition:

Play Video
HWAG

Audiometry is the measurement of sound thresholds at different frequencies. Here, the different plot symbols indicate whether or not a test tone was heard. Tones adaptively zero in on the threshold at one frequency before moving on to the next. Test batteries of any sort generally employ a similar approach. The test battery proceeds one test at a time, and information is not shared across tests. This type of testing has little or no capacity to improve in efficiency or predictive value.

Example of MAT as a more efficient method of testing that shares information across tests:

Play Video
MLAG

Machine learning audiometry enables information to be shared across all the input variables. Here, the conventional audiometric threshold values are superimposed as a line over a fully predictive machine learning model that improves after each new tone. A faster and more informative test results.

Reference

Song XD, Wallace BM, Gardner JR, Ledbetter NM, Weinberger KQ, Barbour DL. “Fast, continuous audiogram estimation using machine learning.” Ear and Hearing, 2015 Nov-Dec;36(6):e326-35, PMID: 26258575, PMCID: PMC4709018, DOI: 10.1097/AUD.0000000000000186

Spread the Word
Back To Top