Related Experiment Videos
Statistical models for predicting Edgerton-Danhauer NST scores from pure-tone thresholds
J L Danhauer1, T L Sahley, C Abdala
1Department of Speech and Hearing Sciences, University of California Santa Barbara 93106.
The Journal of Auditory Research
|July 1, 1986
Summary
Predicting speech recognition scores using pure-tone thresholds (PTTs) is feasible. This study found that PTTs can accurately predict Nonsense Syllable Test (NST) scores, aiding in the assessment of hearing-impaired individuals.
Area of Science:
- Audiology
- Speech-Language Pathology
- Hearing Science
Background:
- Assessing speech recognition is crucial for understanding hearing impairment.
- Pure-tone thresholds (PTTs) are standard audiological measures.
- The Nonsense Syllable Test (NST) is a common tool for evaluating speech recognition, particularly in individuals with sensorineural hearing loss.
Purpose of the Study:
- To determine if PTTs can accurately predict phonemic-scored NST results.
- To identify the most effective PTT combinations for predicting NST scores across various presentation levels.
- To explore the potential of PTT-based prediction models for difficult-to-test populations.
Main Methods:
- Utilized data from 97 participants across four previous studies using consistent stimuli and procedures.
- Analyzed PTTs at octave frequencies (0.25-8 kHz) and weighted combinations.
- Employed stepwise multiple linear regression to develop predictive equations for NST scores at different sensation levels (SLs).
Main Results:
- Pure-tone threshold at 2 kHz (PTT2 kHz) was the single most predictive factor for NST scores.
- Predictive models varied in effectiveness across different NST SLs.
- The most accurate prediction equations at each NST SL often included weighted PTT averages (0.5, 1, and 2 kHz), particularly at higher sensation levels.
- 86-91% of NST scores were predicted within +/- 10% accuracy.
Conclusions:
- PTTs can predict NST scores with significant accuracy.
- Developed models offer a promising method for estimating speech recognition abilities, especially for individuals who cannot be directly tested.
- Further validation on diverse populations may enhance the clinical utility of these predictive equations.