Aided Cortical Auditory Evoked Potentials in Infants With Frequency-Specific Synthetic Speech Stimuli: Sensitivity, Repeatability, and Feasibility.

OBJECTIVES: The cortical auditory evoked potential (CAEP) test is a candidate for supplementing clinical practice for infant hearing aid users and others who are not developmentally ready for behavioral testing. Sensitivity of the test for given sensation levels (SLs) has been reported to some degree, but further data are needed from large numbers of infants within the target age range, including repeat data where CAEPs were not detected initially. This study aims to assess sensitivity, repeatability, acceptability, and feasibility of CAEPs as a clinical measure of aided audibility in infants. DESIGN: One hundred and three infant hearing aid users were recruited from 53 pediatric audiology centers across the UK. Infants underwent aided CAEP testing at age 3 to 7 months to a mid-frequency (MF) and (mid-)high-frequency (HF) synthetic speech stimulus. CAEP testing was repeated within 7 days. When developmentally ready (aged 7-21 months), the infants underwent aided behavioral hearing testing using the same stimuli, to estimate the decibel (dB) SL (i.e., level above threshold) of those stimuli when presented at the CAEP test sessions. Percentage of CAEP detections for different dB SLs are reported using an objective detection method (Hotellings T 2 ). Acceptability was assessed using caregiver interviews and a questionnaire, and feasibility by recording test duration and completion rate. RESULTS: The overall sensitivity for a single CAEP test when the stimuli were ≥0 dB SL (i.e., audible) was 70% for the MF stimulus and 54% for the HF stimulus. After repeat testing, this increased to 84% and 72%, respectively. For SL >10 dB, the respective MF and HF test sensitivities were 80% and 60% for a single test, increasing to 94% and 79% for the two tests combined. Clinical feasibility was demonstrated by an excellent >99% completion rate, and acceptable median test duration of 24 minutes, including preparation time. Caregivers reported overall positive experiences of the test. CONCLUSIONS: By addressing the clinical need to provide data in the target age group at different SLs, we have demonstrated that aided CAEP testing can supplement existing clinical practice when infants with hearing loss are not developmentally ready for traditional behavioral assessment. Repeat testing is valuable to increase test sensitivity. For clinical application, it is important to be aware of CAEP response variability in this age group.

Auditory Brainstem Response Detection Using Machine Learning: A Comparison With Statistical Detection Methods.

OBJECTIVES: The primary objective of this study was to train and test machine learning algorithms to be able to detect accurately whether EEG data contains an auditory brainstem response (ABR) or not and recommend suitable machine learning methods. In addition, the performance of the best machine learning algorithm was compared with that of prominent statistical detection methods. DESIGN: Four machine learning algorithms were trained and evaluated using nested k-fold cross-validation: a random forest, a convolutional long short-term memory network, a stacked ensemble, and a multilayer perceptron. The best method was evaluated on a separate test set and compared with conventional detection methods: Fsp, Fmp, q-sample uniform scores test, and Hotelling's T2 test. The models were trained and tested on simulated data that were generated based on recorded ABRs collected from 12 normal-hearing participants and no-stimulus EEG data from 15 participants. Simulation allowed the ground truth of the data ("response present" or "response absent") to be known. RESULTS: The sensitivity of the best machine learning algorithm, a stacked ensemble, was significantly greater than that of the conventional detection methods evaluated. The stacked ensemble, evaluated using a bootstrap approach, consistently achieved a high and stable level of specificity across ensemble sizes. CONCLUSIONS: The stacked ensemble model presented was more effective than conventional statistical ABR detection methods and the alternative machine learning approaches tested. The stacked ensemble detection method may have potential both in automated ABR screening devices as well as in evoked potential software, assisting clinicians in making decisions regarding a patient's ABR threshold. Further assessment of the model's generalizability using a large cohort of subject recorded data, including participants of different ages and hearing status, is a recommended next step.

Envelope frequency following responses are stronger for high-pass than low-pass filtered vowels.

BACKGROUND: To assess hearing in response to speech, the envelope frequency following response (FFR) can be observed at the fundamental frequency of a vowel stimulus and its harmonics. FFRs are complex non-linear phenomena, which require better understanding for allowing robust inferences on the assessment of hearing and hearing aid fitting. OBJECTIVES: To evaluate the effect of stimulus bandwidth on FFR detection rates using filtered vowel stimuli with equal sound levels. DESIGN: FFRs were collected whilst presenting repeated vowels (in consonant-vowel-consonant format) filtered into different bandwidths. Eighty stimuli per word were presented at 70 dB SPL LAeq through insert earphones with an inter-stimulus interval of 1 s. Responses were detected using frequency-domain Hotelling's T2 (HT2) tests for individual multiples of the fundamental frequency (F0) and for combinations of F0 multiples. STUDY SAMPLE: A total of 11 native English-speaking subjects with normal hearing thresholds. RESULTS: Average detection rates are highest (69%) with stimuli high-pass filtered >1000 Hz, and significantly lower for low-pass filtered stimuli (40%). CONCLUSIONS: High-pass filtered vowels therefore appear to elicit stronger FFRs than low-pass filtered vowels at the same dB SPL LAeq. For testing hearing using band-limited speech, filtering effects (due to hearing loss, hearing aid setting or stimulus choice) on responses must be considered.

Improved Detection of Vowel Envelope Frequency Following Responses Using Hotelling's T2 Analysis.

OBJECTIVES: Objective detection of brainstem responses to natural speech stimuli is an important tool for the evaluation of hearing aid fitting, especially in people who may not be able to respond reliably in behavioral tests. Of particular interest is the envelope frequency following response (eFFR), which refers to the EEG response at the stimulus' fundamental frequency (and its harmonics), and here in particular to the response to natural spoken vowel sounds. This article introduces the frequency-domain Hotelling's T (HT2) method for eFFR detection. This method was compared, in terms of sensitivity in detecting eFFRs at the fundamental frequency (HT2_F0), to two different single-channel frequency domain methods (F test on Fourier analyzer (FA) amplitude spectra [FA-F-Test] and magnitude-squared coherence [MSC]) in detecting envelope following responses to natural vowel stimuli in simulated data and EEG data from normal-hearing subjects. Sensitivity was assessed based on the number of detections and the time needed to detect a response for a false-positive rate of 5%. The study also explored whether a single-channel, multifrequency HT2 (HT2_3F) and a multichannel, multifrequency HT2 (HT2_MC) could further improve response detection. DESIGN: Four repeated words were presented sequentially at 70 dB SPL LAeq through ER-2 insert earphones. The stimuli consisted of a prolonged vowel in a /hVd/ structure (where V represents different vowel sounds). Each stimulus was presented over 440 sweeps (220 condensation and 220 rarefaction). EEG data were collected from 12 normal-hearing adult participants. After preprocessing and artifact removal, eFFR detection was compared between the algorithms. For the simulation study, simulated EEG signals were generated by adding random noise at multiple signal to noise ratios (SNRs; 0 to -60dB) to the auditory stimuli as well as to a single sinusoid at the fluctuating and flattened fundamental frequency (f0). For each SNR, 1000 sets of 440 simulated epochs were generated. Performance of the algorithms was assessed based on the number of sets for which a response could be detected at each SNR. RESULTS: In simulation studies, HT2_3F significantly outperformed the other algorithms when detecting a vowel stimulus in noise. For simulations containing responses only at a single frequency, HT2_3F performs worse compared with other approaches applied in this study as the additional frequencies included do not contain additional information. For recorded EEG data, HT2_MC showed a significantly higher response detection rate compared with MSC and FA-F-Test. Both HT2_MC and HT2_F0 also showed a significant reduction in detection time compared with the FA-F-Test algorithm. Comparisons between different electrode locations confirmed a higher number of detections for electrodes close to Cz compared to more peripheral locations. CONCLUSION: The HT2 method is more sensitive than FA-F-Test and MSC in detecting responses to complex stimuli because it allows detection of multiple frequencies (HT2_F3) and multiple EEG channels (HT2_MC) simultaneously. This effect was shown in simulation studies for HT2_3F and in EEG data for the HT2_MC algorithm. The spread in detection time across subjects is also lower for the HT2 algorithm, with decision on the presence of an eFFR possible within 5 min.