Ajuste do Nível de Significância Associado ao Número de Detecções Consecutivas Aplicada a Detecção de Respostas Auditivas em Regime Permanente

Abstract

The strategy for sequential tests is an option to identify the Auditory Steady-State Responses (ASSRs) faster when compared to the single test. However, the repetitive application of objective response detectors (ORD) increases the probability of detecting a response when it does not exist, i.e., false positive (FP). A solution to this increase in FP is to define the response as present when it reaches a predetermined number of consecutive detections (NDC). However, the NDC can lead to a more conservative exam, which can lead to loss of detection rates. In this work, a method for adjusting the significance level associated with the NDC was proposed, with the objective of maintaining the FP rate of the exam equal to the desired significance level. The proposed method was applied to the EEG signal set composed of 11 volunteers for intensities of 50, 40 and 30 dB SPL. The results demonstrate that the proposed adjustment kept the FP rate of the exam around the desired significance level of 5%. In addition, it obtained a percentage increase in detection rates of 9.68%, 21.23% and 33.20% for the intensities of 50, 40 and 30 dB SPL respectively, as well as a reduction in the examination time when compared to the strategy without the proposed adjustment.