According to the World Health Organization, 466 million people worldwide have disabling hearing loss, and this number will almost double by 2050, i.e. one person every ten people. Despite the significant burden of hearing loss, it is frequently under-diagnosed, particularly in older adults. Hearing ability is evaluated by means of hearing tests. Although Pure-tone audiometry (PTA) is the most diffused hearing test, it is not able to evaluate how much a subject is able to recognize speech stimuli in noise. In this regard, speech-in-noise tests (SNTs) are the most adequate indicators of the effective hearing ability in everyday-like situations. Particularly, in the past few years, interest has been growing in the area of automated SNTs for self-administered hearing testing which might lead to an increase in hearing screening rate. The majority of speech-in-noise tests are based on words, sentences or digit sequences, and they are, consequently, fully dependent on listeners' education and native language. In this regard, the use of Vowel-Consonant-Vowel (VCV) stimuli has become popular for testing adults. VCVs have no meaning and, as such, they do not involve any cognitive recognition processes and are less influenced by the subjects’ education, literacy, and native language. A popular procedure for automated SNTs is the adaptive staircase, i.e. a procedure where the presentation level of test stimuli is adjusted depending on the subject’s responses. However, conventional staircase procedures adjust the stimulus presentation level by using fixed steps, on the assumption that stimuli in the set are equally intelligible, which is not the case for speech stimuli, including VCVs. This study aims at designing a novel fast, accurate and repeatable staircase procedure based on VCV stimuli for speech-in-noise test. The proposed staircase takes into account differences in intelligibility across the set by using changes in stimulus presentation levels that vary depending on the stimulus and the presentation level. Preliminary evaluation of the proposed staircase, as compared to a conventional one, demonstrated excellent test-retest reliability, with very good agreement with the conventional method in terms of speech-in-noise threshold estimation, and significantly lower test duration (up to about two minutes shorter). The estimated robustness in noisy settings resulted higher for the novel method in a comparison between professional and domestic-like test settings as well. As such, the new method shows promise for rapid and reliable self-screening in adults. Further research is needed to assess test variability due to usage of different devices for screening and to evaluate test performance in subjects with hearing loss.
Secondo l'Organizzazione Mondiale della Sanità, 466 milioni di persone nel mondo soffrono di perdita uditiva invalidante e questo numero diventerà quasi il doppio entro il 2050, cioè una persona su dieci. Nonostante la perdita d'udito sia un ingente problema, è frequentemente sotto-diagnosticata, in particolar modo negli adulti e negli anziani. Il livello di udito è misurato attraverso i test dell'udito. Sebbene l'audiometria tonale (Pure Tone Audiometry, PTA) sia il test dell'udito più diffuso, non è in grado di misurare l'effettiva abilità del soggetto nel riconoscere stimoli vocali immersi nel rumore. In tal senso, i test speech-in-noise (SNT) ("parlato nel rumore") risultano i migliori indicatori dell'effettivo livello uditivo in situazioni quotidiane. In particolar modo, negli anni recenti, è cresciuto l'interesse nell'area degli SNT per auto somministrazione di test dell'udito, che potrebbero portare ad un aumento nel tasso di screening uditivo. La maggior parte dei test speech-in-noise si basano sull'uso di parole, frasi o sequenze di numeri e sono, di conseguenza, fortemente dipendenti dall'educazione e dalla lingua nativa dell'ascoltatore. A tal proposito, l'uso di stimoli del tipo Vocale-Consonante-Vocale (VCV), è diventato comune nel testare gli adulti. Le VCV non hanno un significato e, pertanto, non implicano alcun processo di riconoscimento cognitivo e sono meno influenzate dal livello di istruzione, alfabetismo e dalla lingua nativa dei soggetti. Un metodo comune per i test SNT è la adaptive staircase ("scala adattativa"), ovvero un metodo in cui il livello di presentazione degli stimoli del test è regolato a seconda delle risposte del soggetto. Le procedure staircase convenzionali aggiustano il livello di presentazione dello stimolo, mediante l'implicita assunzione che gli stimoli nel set siano egualmente intelligibili. Questo però non è il caso per gli stimoli vocali, incluse le VCV. Questo lavoro mira a progettare una nuovo metodo staircase per gli speech-in-noise test basato sulle VCV. La staircase proposta prende in considerazione differenze in intelligibilità tra gli stimoli, grazie a variazioni nei livelli di presentazione dello stimolo che cambiano a seconda dello stimolo e del livello di presentazione. Una valutazione preliminare della staircase proposta, confrontata con una convenzionale, ha dimostrato un'eccellente affidabilità tra test e retest, grande concordanza con il metodo convenzionale in termini di stima della soglia uditiva e una durata significativamente inferiore di test (circa due minuti in meno). La robustezza delle misure stimata in diverse condizioni di rumore, è risultata superiore nel nuovo metodo nel confronto tra condizioni di test professionali e domestiche. Pertanto, il nuovo metodo sembra essere promettente per una rapida e affidabile auto-somministrazione di test dell'udito negli adulti. Ulteriori ricerche sono necessarie per verificare la variabilità dovuta a uso di dispositivi diversi per lo screening e per valutare l'efficienza del test in soggetti con perdite uditive.
Development and evaluation of a novel adaptive staircase procedure for automated speech-in-noise testing
POLO, EDOARDO MARIA;ZANET, MARCO
2017/2018
Abstract
According to the World Health Organization, 466 million people worldwide have disabling hearing loss, and this number will almost double by 2050, i.e. one person every ten people. Despite the significant burden of hearing loss, it is frequently under-diagnosed, particularly in older adults. Hearing ability is evaluated by means of hearing tests. Although Pure-tone audiometry (PTA) is the most diffused hearing test, it is not able to evaluate how much a subject is able to recognize speech stimuli in noise. In this regard, speech-in-noise tests (SNTs) are the most adequate indicators of the effective hearing ability in everyday-like situations. Particularly, in the past few years, interest has been growing in the area of automated SNTs for self-administered hearing testing which might lead to an increase in hearing screening rate. The majority of speech-in-noise tests are based on words, sentences or digit sequences, and they are, consequently, fully dependent on listeners' education and native language. In this regard, the use of Vowel-Consonant-Vowel (VCV) stimuli has become popular for testing adults. VCVs have no meaning and, as such, they do not involve any cognitive recognition processes and are less influenced by the subjects’ education, literacy, and native language. A popular procedure for automated SNTs is the adaptive staircase, i.e. a procedure where the presentation level of test stimuli is adjusted depending on the subject’s responses. However, conventional staircase procedures adjust the stimulus presentation level by using fixed steps, on the assumption that stimuli in the set are equally intelligible, which is not the case for speech stimuli, including VCVs. This study aims at designing a novel fast, accurate and repeatable staircase procedure based on VCV stimuli for speech-in-noise test. The proposed staircase takes into account differences in intelligibility across the set by using changes in stimulus presentation levels that vary depending on the stimulus and the presentation level. Preliminary evaluation of the proposed staircase, as compared to a conventional one, demonstrated excellent test-retest reliability, with very good agreement with the conventional method in terms of speech-in-noise threshold estimation, and significantly lower test duration (up to about two minutes shorter). The estimated robustness in noisy settings resulted higher for the novel method in a comparison between professional and domestic-like test settings as well. As such, the new method shows promise for rapid and reliable self-screening in adults. Further research is needed to assess test variability due to usage of different devices for screening and to evaluate test performance in subjects with hearing loss.File | Dimensione | Formato | |
---|---|---|---|
2019_04_Polo_Zanet.pdf
Open Access dal 05/04/2020
Descrizione: Testo della tesi
Dimensione
4.25 MB
Formato
Adobe PDF
|
4.25 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/146105