Efficient Adaptive Speech Reception Threshold Measurements Using Stochastic Approximation Algorithms.
Gertjan Dingemanse, André Goedegebure
PMID 32425135WHAT IT FOUND
Simulations show word-scoring adaptive tests are more accurate than standard sentence-scoring methods for hearing-in-noise thresholds, especially in cochlear implant users.
Clinicians should use at least 20 sentences and start the test below the patient's estimated threshold to avoid errors.
Key findings
01Stochastic Approximation (SA) algorithms using word scoring provided more efficient estimates of the speech reception threshold (SRT) than clinically used adaptive procedures in cochlear implant users.
02The averaged SA algorithm with a step size constant of 4 dB is recommended for clinical use because it is simple and works well across a wide range of hearing-impaired listeners.
03For reliable estimation in cochlear implant users, the test must start at an SNR below the true threshold (by 2-4 dB) and use at least 20 sentences; starting too high or using fewer sentences causes significant bias and error.
STILL TO COME
How it was doneWhat they foundWhat it means for SLPs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study is a simulation, not a clinical trial with new patient data. It relies on a model of speech recognition fitted to a small group of participants (20 CI users, 16 NH listeners). The model used VU sentences, which are open-set. The authors note that results may differ for closed-set matrix tests, although they did run some preliminary simulations for those. The simulation assumed a constant lapse rate (inattention) of 2%, which may not reflect real-world variability in patient attention or fatigue. The model was based on relatively high-performing CI users, so results may not apply to those with poorer speech recognition skills.
Declared interests
The authors declare no conflicts of interest. The study was supported by the Netherlands Organisation for Scientific Research (NWO).
The easy way to misread this
Do not assume that a low 'within-staircase SD' during the test means the final threshold estimate is accurate. The study found no relationship between this within-test variability and the actual accuracy of the threshold. Rely on the algorithm's parameters (starting SNR, list length) and the patient's quiet scores instead.