Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Sep 8.
Published in final edited form as: Ear Hear. 2025 Oct 13;47(2):470–479. doi: 10.1097/AUD.0000000000001744

Electrophysiological responses to spectral ripple envelope phase inversion in 3-month-olds and adults

Talat Jabeen 1,2,*, David Horn 1,2,*, Mark W Pettet 3, Emily A Hammond 1, Jay T Rubinstein 1,2, Bonnie K Lau 1,2,3,4
PMCID: PMC13549533  NIHMSID: NIHMS2153089  PMID: 41077645

Abstract

Objectives:

A critical problem in the clinical management of infants is the limited methods available to assess whether hearing aids or cochlear implants are providing sufficient input for speech sound discrimination. Spectral ripple discrimination is one non-linguistic, behavioral measure that has been shown to correlate with speech perception in both adults and children. Electrophysiological responses to spectral ripple stimuli have been proposed as a potentially clinically feasible tool. In this study, a behavioral spectral ripple envelope inversion discrimination paradigm was adapted to investigate whether an electrophysiological response can be obtained using electroencephalography (EEG) in infants. Specifically, we investigate whether the detection of envelope inversion via the acoustic change complex (ACC) would match previously reported behavioral discrimination of the same stimuli.

Design:

Spectral ripple stimuli included two ripple densities selected based on prior behavioral results: 1 ripple-per-octave (RPO), which can be discriminated by both infants and adults, and 8 RPO, which was not discriminated by either age group. Stimuli were constructed from two concatenated 1-s noise carriers, with the spectral envelope sinusoidally modulated at either 1 RPO or 8 RPO with 10 dB depth. To elicit an ACC, a spectral envelope inversion was imposed mid-stimulus at 1 second for both RPOs. Adult and 3-month-old infant participants were presented with stimuli from both conditions while undergoing 32-channel EEG recording.

Results:

At group level, an ACC to the spectral envelope inversion was detected only in the 1 RPO stimulus condition but not the 8 RPO condition for both infant and adult groups. In contrast, an Onset response to the stimuli was detected in both conditions, demonstrating that the stimuli themselves were detected. In the 1 RPO condition, the ACC detection rate was 81% for individual infants and 73% for individual adults while in the 8 RPO condition, an ACC was detected in one infant and no adult participants. Individual participant detection rates for the Onset response were 94% for infants and 100% for adults in both conditions.

Conclusions:

These findings show a relationship between EEG ACC detection and previously reported behavioral discrimination abilities, with the ACC detected in the 1 RPO but not 8 RPO condition. These results demonstrate the ACC’s potential for infant discrimination threshold estimation because it was elicited by a suprathreshold acoustic change (1 RPO) but not a subthreshold acoustic change (8 RPO). Critically, both 1 RPO and 8 RPO stimuli contained an envelope inversion at 1s, suggesting the ACC could serve as an objective measure of stimulus discriminability in infants as young as 3 months. Future studies with more stimulus conditions to fully establish a threshold, online monitoring of responses to increase success rate of infant testing, and comparison of testing infants awake versus asleep are needed to further develop this approach.

Keywords: Electrophysiology, Infant Assessment, Spectral Ripple Envelope Inversion, Acoustic Change Complex

Introduction

Early establishment of auditory input and habilitation is crucial for optimal developmental outcomes (Korver et al., 2010; Yoshinaga-Itano et al., 1998, 2018). Newborn hearing screening has enabled the identification of hearing loss within the first few months of life. However, early receipt of hearing technology such as hearing aids (HA) or cochlear implants (CI) does not always lead to optimal speech perception and language outcomes (Ching et al., 2017; Moeller et al., 2007; Niparko et al., 2010; Tomblin et al., 2015). One factor that may contribute to this variability in outcomes, is whether a child’s hearing device is providing sufficient input for them to discriminate speech sounds. When children reach 4–5 years of age, assessment of device efficacy can be conducted in the clinic using measures of speech recognition (Eisenberg et al., 2006; Uhler et al., 2017). However, even in children with typical neurocognitive function and early implantation, tremendous variability in clinical outcomes are apparent by this age (Niparko et al., 2010). The inability to obtain precise measures of speech perception in infants makes it difficult to optimize auditory input via CI mapping or HA settings. It is also difficult to track the development of speech perception to guide language habilitation decisions (Culbertson et al., 2022). While some work has been done to develop basic methods to assess speech sound discrimination, robust clinically feasible and relevant measures have not yet emerged (Cone-Wesson, 2003; Eisenberg et al., 2007; Martinez et al., 2008; Uhler et al., 2011). Moreover, these methods have typically focused on English speech sounds and may not be applicable to infants with varied linguistic environments.

The acoustic change complex (ACC) is an auditory evoked potential elicited by an acoustic change in an ongoing sound (Martin & Boothroyd, 1999). There is a large body of research showing that the ACC can be elicited to numerous types of acoustic changes in adult listeners, including amplitude modulation (Han & Dimitrijevic, 2015), frequency and intensity changes (e.g., Dimitrijevic et al., 2008; He et al., 2012; Kalaiah, 2018), and acoustic changes in speech sounds (e.g., Cheek & Cone, 2020; Martin et al., 2010; Ostroff et al., 1998). Importantly, it can also be elicited in CI and HA users (Biot et al., 2024; Brown et al., 2008; Friesen & Tremblay, 2006; Liang et al., 2018; Tremblay, Billings, et al., 2006; Tremblay, Kalstein, et al., 2006; Undurraga et al., 2021). The ACC has been identified as a good candidate for objective evaluation of speech perception abilities as it can be elicited with relatively few stimulus presentations, has good test-retest reliability, shows agreement with behavioral measures, and can be recorded in hard of hearing individuals (reviewed in Kim, 2015).

Early infant studies conducted by Small and colleagues illustrated that eliciting infant ACC requires special consideration of stimulus parameters to account for neural immaturity in the infant auditory system (Chen & Small, 2015; Small & Werker, 2012). They showed that when longer duration consonant-vowel speech tokens were presented, ACCs were more reliably elicited and showed more robust morphology than with shorter stimulus durations. While there are comparatively few infant ACC studies, several others since then have recorded infant ACCs to consonant and vowel contrasts (Ching et al., 2023; Cone et al., 2022; Martinez et al., 2013; Strahm et al., 2022; Uhler et al., 2018), iterated ripple noise (Strahm et al., 2022) and high-pass filtered spectral ripple noise (Ching et al., 2023). These findings suggest that the ACC can indeed be reliably elicited in infants across the first year of life to a variety of acoustic parameters. Nonetheless, Uhler and colleagues found that for sleeping infants between 1.5 and 5 months of age, the mismatch response (reviewed in Näätänen et al., 2007) was more reliably elicited to vowel changes than the ACC, despite both reflecting acoustic change detection. However, the interpretation of the mismatch response in individual infants is difficult. Together, these findings suggest that while infant ACCs can be elicited, many aspects of the optimal study design, testing conditions, and stimulus parameters require further systematic investigation.

More recent studies demonstrate the clinical applicability of the ACC for infant assessment. Ching and colleagues tested a large sample of 43 infants with typical hearing and 72 who are hard of hearing (HoH) between 3 and 12 months of age. They showed that the ACC detection rate of aided infants with mild or moderate hearing loss was no different than typical hearing infants, although those with a severe hearing loss showed a lower rate of detection (Ching et al., 2023). This finding demonstrates that the ACC can indeed be recorded in aided infants, and thus, has the potential to serve as a measure of device efficacy. Furthermore, Cone and colleagues, found that ACCs to vowel contrasts could be recorded in some infants before they showed behavioral discrimination of the same contrasts (Cone et al., 2022). Since behavioral testing is challenging in younger infants, the ACC may a useful alternative for both typical hearing and hard of hearing infants.

As noted earlier, non-linguistic measures may be more optimal than speech sound contrasts for assessing very young patients with different language environments. Spectral resolution, the ability to perceive the pattern of sound intensity variation across frequency, is important for speech understanding, especially in complex acoustic environments. Spectral ripple discrimination is one method that has been used to measure spectral resolution in typical hearing listeners as well as those who are hard of hearing (An et al., 2022; Brennan & McCreery, 2021; Nechaev et al., 2024). For adults as well as children, spectral ripple discrimination has been shown to correlate with speech perception (Anderson et al., 2011; DiNino & Arenberg, 2018; Henry et al., 2005; Horn, Dudley, et al., 2017; Noble et al., 2024; Won et al., 2007). Moreover, using CI vocoder simulations, Won and colleagues showed that the ACC recorded to vocoded spectral ripple stimuli correlated to discrimination of the stimuli measured behaviorally (Won et al., 2011). Past studies also show that behavioral measures of spectral ripple discrimination have been successfully obtained in infants (Horn, Won, et al., 2017; Noble et al., 2023). While obtaining a psychophysical threshold in infants is possible, it is often time consuming and requires multiple visits to the laboratory (e.g., Horn, Won, et al., 2017; Lau et al., 2021). Thus, this study investigates whether an electrophysiological measure of discrimination can be obtained in infants.

While spectral ripple discrimination has been widely used in studies of spectral resolution, performance is affected by both resolution of spectral peak location (frequency resolution) and sensitivity to intensity differences across spectrum (Horn, Won, et al., 2017; Noble et al., 2024; Supin et al., 1999). Sensitivity to intensity differences is immature in infants and could confound results of any EEG-based spectral ripple discrimination paradigm. Additionally, non-spectral intensity cues can arise with spectral ripple stimuli, particularly when the ripple densities approach the limits of peripheral spectral resolution (Anderson et al., 2012; Winn & O’Brien, 2022). While this can be addressed by using shallow ripple depths (hence avoiding higher ripple density thresholds), other spectral resolution paradigms may be more robust to these confounding non-spectral cues (Aronoff & Landsberger, 2013).

The ultimate goal of this research is to develop and validate an objective measure of hearing device efficacy that can guide clinical management at an earlier age than is currently possible. One benefit of a non-linguistic test of acoustic perception is that it does not require any prior linguistic knowledge or competence, making it appropriate for use with preverbal infants as well as culturally and linguistically diverse populations. The use of EEG as a method is also advantageous because it is affordable and clinically scalable. Moreover, EEG protocols can be maintained across age groups. This allows infants, children, and adults can be tested with the same method, making comparison of results and clinical implementation more straightforward.

While past studies have shown that infant ACCs can be elicited to acoustic change when stimulus parameters are designed for infant recording, it is unknown whether the ACC can be used to establish a discrimination threshold. To determine a threshold, it must be clear which stimuli elicits an ACC and which stimuli do not, for each individual infant tested (i.e., not group level analysis). In the current study, participants included 3-month-old infants in addition to a group of adults to serve as mature comparison participants. We tested 3-month-olds because based on the 1–2-3 rule on infant hearing, the goal is to fit hearing aids by 3 months (Joint Committee on Infant Hearing, 2019). One of the challenges of testing young infants awake is that the protocol must be short because they have limited durations of wakefulness. Thus, we selected only two spectral ripple stimulus conditions. One ripple-per-octave (RPO), which was discriminated by both infants and adults and 8 RPO, which was not discriminated by either age groups in prior behavioral studies using the same stimuli (Horn, Won, et al., 2017). We ask the question: Can the ACC be elicited by some acoustic changes but not others, based on stimulus discriminability? We hypothesized that an ACC will be elicited in the 1 RPO condition but not the 8 RPO condition, and thus, show potential to serve as an objective measure of discrimination.

Methods

Participants

Participants in this study were 24 3-month-old infants (mean:13 weeks, range: 11–14 weeks, 16 female) and 11 adults (mean: 21 years, range: 18–26 years, 6 female). All infant participants were born full term, had no history of otitis media within 2 weeks of testing, passed their newborn hearing screening and had no history of health or developmental concerns. All adult participants had no history of hearing loss, otitis media within 2 weeks of testing, and had no health or neurological concerns that would interfere with testing. Race was self-reported by adult participants and included 7 participants identifying as Asian, 1 as Black, and 3 as More Than One Race. Infants’ race and ethnicity were based on parent report, including 3 participants reported as Asian, 3 as More Than One Race, 17 as White, and 1 Unknown. Data obtained from 8 infants was excluded from analysis due to infant falling asleep (2), infant being too fussy to complete testing (4), excessive movement during recording (1) and experimental error (1). These stringent data exclusion criteria were adopted to ensure that data from sleeping or sleepy infants were excluded. All participants were tested in a single visit lasting approximately one hour. Infant participants were required to pass a tympanometric screen using a 1000 Hz probe tone to ensure that they had no fluid in their ears. Adult participants were required to pass an audiometric screen (<20 dB HL at octave frequencies between 500 Hz to 8 kHz) at the time of testing. All methods were performed in accordance with protocols approved by the Institutional Review Board at the university where the research was conducted. Written informed consent was obtained from adult participants or parent/guardians of infant participants, and all participants were provided with monetary compensation for their time.

Stimuli

The stimuli were constructed using a custom MATLAB script based on prior behavioral studies so that behavioral and electrophysiological findings can be compared (Fig.1; Horn, Won, et al., 2017; Won et al., 2011). The spectral ripple stimulus was constructed by concatenating two 1-second noise carriers, matched in root-mean-square amplitude, consisting of 2555 random phase pure tones between 100–5000 Hz. The overall duration of the stimulus was 2 seconds with 150 ms rise fall linear ramps. The spectral envelope was sinusoidally modulated at either 1- or 8-RPO with peak-to-trough depth of 10 dB, corresponding to the two conditions in this study. Both typical hearing and cochlear implant users can sometimes discriminate very high ripple densities at depths greater than 15–20 dB, which are attributed to spurious non-spectral temporal cues (Horn, Dudley, et al., 2017; Winn & O’Brien, 2022). To reduce the likelihood that non-spectral cues would be perceived, a ripple depth of 10 dB was selected. The stimulus was then filtered with a long-term speech-spectrum-shaped filter (Byrne et al., 1994) and low-pass filtered at 5 kHz to eliminate high frequency artifact at the point of concatenation. The starting phase of the first half of the stimulus was varied randomly while the starting phase of the second half was “inverted”, or shifted by 90-degrees, relative to the first. For each of the two stimulus conditions (1 or 8 RPO) there were ten stimulus tokens that each had a different starting phase and were randomly selected upon each presentation.

Fig. 1.

Fig. 1.

Open in a new tab

Spectral ripple stimulus with an envelope phase inversion at 1 s (red arrow) for 1 RPO (A) and 8 RPO (B). Both stimuli have an envelope phase inversion at 1s.

EEG Acquisition

Infants were tested awake to elicit the most robust neural response. EEG data were recorded at a sampling rate of 1,000 Hz using a Biosemi EEG system with 32 electrodes placed according to the international 10–20 system, along with two additional electrodes placed on the mastoids for re-referencing. Electrode offsets were kept below 50 mV for both infant and adult participants. During testing, infants sat in their caregiver’s lap inside a sound-attenuated booth with an assistant inside the booth showing the infant toys and books to keep them quiet and attentive. The assistant maintained arousal and attention of the infant during testing to the best of their ability. The experimenter outside the booth was live coding participant arousal, attention, and movement throughout the test session. Adult participants sat alone in the booth. To account for neural immaturity in infants (Small & Werker, 2012), a long interstimulus interval (ISI) of 1.8 s, with a ± 200ms jitter was used. The stimuli were presented at 75 dB SPL via a loudspeaker placed at 0° azimuth 1 meter from the participant. All stimuli were calibrated using a Larsen Davis sound level meter prior to testing. Spectral ripple noise stimuli with randomized starting phase from each of the two RPO conditions were randomly selected and presented. The total recording time was about 16 minutes for 250 total trials, 125 per condition.

EEG Analysis

EEG analysis was conducted using EEGLAB (Delorme & Makeig, 2004) and custom MATLAB (Natick, MA: TheMathWorks Inc.) scripts. The same pre-processing and analysis steps were applied to infant and adult data, including artifact and bad channel removal. Raw data were down sampled to 250 Hz, bandpass filtered (0.5–20 Hz, Hamming FIR), and re-referenced to the average of the mastoid electrodes. Given the prevalence of movement artifacts in infant EEG data, Artifact Subspace Reconstruction (ASR) was applied to remove artifacts caused by transient events (e.g., muscle activity, eye blinks, and movement) as in prior infant EEG studies (e.g., Attaheri et al., 2022; Kalashnikova et al., 2018). ASR adaptively identifies and corrects segments of the data that deviate significantly from a baseline derived from clean EEG epochs. The EEG data is analyzed within sliding time windows and data is rejected based on a threshold that defines how much a channel can deviate from the baseline before being considered an artifact. ASR was implemented with a standard deviation threshold of k = 5 and a sliding window length of 0.5 seconds. Following ASR, remaining channels with excessive noise or abnormal voltage levels were identified by calculating the kurtosis of each channel and the threshold for rejection was set to 2 standard deviations from the mean. The removed channels were then interpolated using spherical interpolation. The additional bad channel removal and interpolation was applied to help retain channel consistency across subjects and conditions. It also allowed for interpretation of the global field power (GFP), which would be affected by the presence of noisy channels. Epochs were then extracted from −100 ms to 2000 ms relative to stimulus onset. Epochs were baseline corrected by subtracting the mean voltage in the −100 ms to 0 ms pre-stimulus period from the entire epoch.

Acoustic Change Complex Identification

Data from the Cz electrode was selected for analysis as it is a midline electrode known to capture neural activity related to sound processing. It is also commonly used in clinical recording preparations. It was a priority in this study to focus on individual participant analysis. Since a single infant electrode trace can be difficult to interpret due to reduced signal-to-noise ratio, we also assessed the GFP to provide additional information from the whole head. The GFP quantifies the overall activity or variability of the recorded signals across the scalp, computed as the standard deviation of the EEG signals across all electrodes. Thus, the GFP provides a summary of overall brain activity with higher GFP values indicating a larger difference between electrode voltages, suggesting stronger or more global brain activation. The GFP also complements single electrode event related potential (ERP) analysis as peaks in the GFP often correspond to significant events in the ERP waveform. This can assist in the identification of components such as the ACC in infant data. Finally, the GFP provides confirmation that our artifact removal pipeline was effective. The Onset and ACC responses were identified automatically with a custom MATLAB function that finds local maxima in the Cz trace and the GFP. Peak detection was constrained to a time window 0.15 to 0.30 s post stimulus onset or acoustic change. Thresholding criteria were also applied to exclude small fluctuations in the signal that did not meet the minimum amplitude criteria of 3.5 μV for the Cz trace and 1.0 μV for the GFP trace. The minimum amplitude criteria were established based on characterization of the noise floor at group level. Note that for adult participants, this time window is targeting the P2 peak, which is more prominent in the adult waveforms. Fitted peaks within the expected time window were then visually confirmed. If a fitted peak was not prominent, the peak was removed. Onset and ACC responses were determined as detected, only if a peak was fit for both Cz and GFP traces. The ACC peak for Adult 9 was fit manually for the Cz trace (denoted with a blue square in Fig. 4), as their response morphology was a negative deflection and not detected by the algorithm. Infant 6 had an onset response outside of the selected time window. However, the onset peak was clear and thus, the time window was not extended (see Discussion section for further considerations).

Fig. 4.

Fig. 4.

Open in a new tab

Individual infant responses with electrode Cz plotted on top and GFP on bottom. In the 1 RPO condition, no onset response was detected for Infant 1 and no ACC was detected for Infant 1, Infant 8, and Infant 15. For Infant 8, an ACC peak was also detected at 8 RPO.

Results

Figure 2 shows data from one individual adult subject, time aligned to the 1 RPO stimulus. Grand average waveforms from electrode Cz as well as the GFP from both conditions are shown in Figure 3. Onset responses are observed in both conditions for both age groups, confirming detection of both 1 RPO and 8 RPO stimuli. Infant Onset response morphology is characterized primarily by a single positive peak, as opposed to the mature P1-N1-P2 pattern seen in adults. At group level, an ACC is elicited in the 1 RPO condition but not the 8 RPO condition for both infants and adults, consistent with our hypothesis. As observed for the Onset response, the ACC in 3-month-olds is also characterized by a single positive peak.

Fig. 2.

Fig. 2.

Open in a new tab

Spectral ripple stimulus (A, 1 RPO) time aligned with the neural response of one individual adult participant (B). Red dotted lines mark the onset of the stimulus (0s) and the envelope inversion (1s).

Fig. 3.

Fig. 3.

Open in a new tab

Grand averaged responses with 1 RPO in blue and 8 RPO in orange. Infant Cz electrode shown in (A) and GFP in (C). Adult Cz electrode shown in (B) and GFP in (D). Onset and ACC responses are labeled in A, and the spectral ripple inversion which would elicit an ACC is mark with a dotted line at 1s. The circle markers show the peaks fitted by the custom MATLAB algorithm used for response detection.

Individual data for infant participants are shown in Figure 4 and individual data for adult participants are shown in Figure 5. Figure 6 shows latencies and amplitudes of peaks extracted by the custom MATLAB algorithm used for response detection. Again, note that for adult participants, these amplitude and latencies are for the more prominent P2 peak. In the 1 RPO condition, an ACC was detected in 13 of 16 or 81% of infants and in the 8 RPO condition, an ACC was detected in 1 of 16 or 0.06 % of infants. In the 1 RPO condition, an ACC was detected in 8 of 11 or 73% of adults while in the 8 RPO condition, no ACCs were detected. In contrast, an Onset response was detected in all but one infant in both conditions equating to a 97% detection rate. For adults, an Onset response was detected in all participants in both conditions. Hearing screening via otoacoustic emissions or auditory brainstem responses was not repeated at the time of testing since the Onset response provides evidence that participants detected the stimuli. However, in the case of the infant that did not show an Onset response, it is possible that hearing loss could have affected the recorded neural response.

Fig. 5.

Fig. 5.

Open in a new tab

Individual adult responses with electrode Cz plotted on top and GFP on bottom. An onset response was detected for all adult participants, but ACC was not detected at 1 RPO for Adult 2, Adult 3, and Adult 11.

Fig. 6.

Fig. 6.

Open in a new tab

Onset and ACC Latency and Amplitude (mean ± 1SEM by age group, 1 RPO in blue, 8 RPO in orange). Onset response latencies are shown in A, Onset response amplitudes are shown in B, ACC response latencies are shown in C and ACC response amplitudes are shown in D for 1 RPO only. Note that for adult listeners, latency and amplitude corresponds to the more prominent P2 peak.

To address the main research question of whether the ACC can be elicited by some acoustic changes but not others, a logistic regression model was used to examine the relationship between condition (1 RPO vs. 8 RPO), age group (infants vs. adults) and response detection for the ACC (present/absent). The coefficients of the logistic regression model were exponentiated to obtain odds ratios (ORs) with 95% confidence intervals (CIs). The 1 RPO condition was associated with a higher odds of detecting an ACC compared to the 8 RPO condition (OR = 2.24, 95% CI [1.94, 2.59], p <0.001). This result was highly significant with a medium effect size. The odds of ACC detection did not differ significantly by age group (OR = 0.93, 95% CI [0.78, 1.11], p=0.423). To further confirm the specificity of the ACC, a second logistic regression was conducted to determine the relationship between condition, age group, and Onset response detection. For the Onset response, no significant effect of condition (OR = 1, 95% CI [0.90, 1.11], p>0.99) or age (OR = 1.06, 95% CI [0.81, 1.13], p=0.245) was observed. Thus, the two logistic regression models confirm that an ACC was detected more often in the 1 RPO condition than the 8 RPO condition but Onset response detection rate was equivalent in both conditions.

Discussion

The aim of this study was to determine whether the ACC is differentially elicited based on stimulus discriminability. This means that for subthreshold stimuli (i.e., RPOs above the listener’s threshold and therefore not perceivable), an ACC should not be detected. That is, for a RPO above the listener’s hearing threshold, meaning it is not consciously perceived. The stimuli presented in this study were identical to those used previously to measure spectral ripple discrimination in infants using an observer-based psychoacoustic paradigm (Horn et al., 2017). The EEG results were consistent with behavioral performance with an ACC elicited to 1 RPO but not 8 RPO stimuli (Horn, Won, et al., 2017). Data from infant and adult participants were robust enough to analyze at the individual level although some variability in detection rates were observed. An ACC was detected in the 1 RPO condition in 73% of adults and 81% of infants and an ACC was detected in the 8 RPO condition in one infant. In contrast, the Onset response was detected in all participants except one infant for both conditions, suggesting that the stimuli themselves were detected. While previous infant studies show successful recordings of the ACC, this is the first infant study to demonstrate its potential for threshold estimation. In our study, an ACC was not elicited by a subthreshold acoustic change (8 RPO). This finding confirms the primary research question that the ACC is a good candidate for an objective measure of stimulus discriminability in infants.

The ACC detection rates in this study were generally comparable to past infant studies. Ching and colleagues (2023) recorded ACCs to a spectral ripple noise with similar parameters to this present study (i.e.,1 RPO, 20 dB depth, 90-degree phase shift as the acoustic change), but reported a lower ACC detection rate of 64% compared to the 81% found in this study. However, their spectral ripple stimulus was high pass filtered at 2000 Hz, possibly explaining the lower detection rate due to the lack of low frequency information in the stimulus. Moreover, the 81% detection rate found in this study is comparable to their speech stimulus conditions, [uiu] and [szs], that did contain low frequency information and yielded a detection rate of 95%. Similarly, Cone (2022) reported a 90% detection rate for infant ACCs elicited by a change in synthesized vowel tokens. Future studies are needed to investigate whether infant ACCs are modulated by stimulus discriminability as they are for adults. For example, Dimitrijevic and colleagues found that the ACC amplitude and latency elicited by spectral changes was more pronounced in the low compared to high frequencies for pure tones (Dimitrijevic et al., 2008). If stimuli closer to threshold produce a lower amplitude ACC in comparison to suprathreshold stimuli, this would be beneficial for establishing discriminability across a range of stimuli. In this present study, since there were only two conditions, one which did not elicit an ACC, modulation of the ACC amplitude and latency could not be assessed.

One challenge of recording EEG in awake infants is the presence of movement artifacts. As data was processed offline, it was possible to include artifact removal and bad channel interpolation in the data processing pipeline. Assessing the GFP in addition to channel Cz was beneficial for confirming the presence of a response. Although we observed differences in the amplitude and latencies of infant and adult responses we defined a single time window and minimum amplitude criterion for both infant and adult data. This is one possible explanation for why an ACC was detected in the 8 RPO condition for Infant 8: As lower amplitudes are accepted across a longer time window the likelihood a false peak is fitted also increases. In general, the automatic peak detection algorithm had a high success rate, except in two cases. Peak detection failed for one adult participant (Adult 9) that showed a deflection instead of a peak and one participant that had a peak outside of the defined time window (Infant 6). These two participants illustrate the importance of refining both peak detection parameters. Fitting both maxima and minima and extending the time window would have resulted in successful peak detection in those cases. Finally, there were 3 adult and 3 infant participants in which an ACC peak was not detected in the 1 RPO condition. It is possible that relaxing the minimum amplitude criteria and/or defining different peak detection time windows may improve the detection rate.

There are several limitations of this study to consider. As auditory brainstem responses were not recorded, it remains possible that infants showing an ACC could still have auditory neuropathy spectrum disorder (i.e., exhibit cortical responses despite brainstem deficit). Additionally, behavioral thresholds were not obtained from participants in this study as one threshold requires 3 to 4 visits to the laboratory on average. It is difficult for most infants to return for so many visits, highlighting the need for more efficient, objective measures to develop clinically feasible tools. Since this study only presented 2 stimulus conditions, one that elicited an ACC and one that did not, behavioral performance is less likely to differ across individual infants than if intermediate stimuli closer to threshold were included. A related limitation is that this study only consisted of 2 conditions because the focus was on very young infants that have limited durations of wakefulness. Thus, one important future direction is refining data acquisition and analysis pipelines to allow for recording more conditions per session to fully establish a threshold. For example, a psychophysical method of constant stimuli approach where more stimuli spanning a range from sub-threshold to suprathreshold would need to be presented. While most of the infant participants had more than 110 usable trials in each condition, Infant 4 and Infant 9 only had ~70–80 trials per condition. Nonetheless, robust Onset and ACCs comparable to the other infant participants were elicited, suggesting that trial numbers can be decreased. However, determining the minimum number of trials required to elicit a response for all infant participants is inherently challenging. It is perhaps an inefficient approach due to the unpredictable occurrence of transient movement artifacts during each recording session. Rather, employing automated online monitoring and averaging of responses during recording may be a preferable option to improve detection rate and shorten recording time for infants.

Future studies should also compare ACCs obtained in infants asleep and awake with the same stimuli. Testing infants asleep may decrease motion artifacts and would enable a longer recording time but it is unknown whether being asleep would attenuate the ACC. One of the biggest challenges to testing very young infants as in this study is managing their duration of wakefulness. To be conservative, all data from sleeping and fussy/sleepy infants were excluded from analysis in this present study. It is important to note nonetheless, that older infants have increased durations of wakefulness and are better at tolerating longer recording durations. Finally, besides adding additional stimulus conditions, another potential future direction is establishing whether responses in a single condition is sufficient to evaluate device efficacy and/or to predict outcomes.

This study employed spectral ripple envelope inversion to elicit the ACC using the same stimuli from a prior infant behavioral study. Behaviorally, infants are able to discriminate the spectral ripple inversion in the 1 RPO condition (Horn, Won, et al., 2017), and in this current study, the ACC was detected in over 80% of the infants tested with this stimulus. In contrast, infants did not show behavioral discrimination of the acoustic change in the 8 RPO condition (Horn, Won, et al., 2017), and consistent with behavioral performance, the ACC was only detected in one infant. These findings demonstrate that not all acoustic changes elicit an ACC and thus, the ACC is a good candidate for an objective measure of discriminability. Further investigations are required to determine whether the ACC can be used to establish an electrophysiological discrimination threshold in infants, which can be adapted to different stimuli.

Financial Disclosures/Conflicts of Interest

This work was supported by NIH R00 DC016640 to B.K.L., a Virginia Merrill Bloedel Hearing Research Center Mini-Grant to B.K.L and D.H. and the UW Auditory Neuroscience Training Grant. NIH T32DC005361-21 to T.J. The authors have no conflicts of interest to declare.

Footnotes

All procedures were conducted according to protocols approved by the Institutional Review Board at the University of Washington (STUDY00006930).

Data Availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author upon request.

References

  1. An YH, Lee ES, Kim DH, Oh HS, Won JH, & Shim HJ (2022). Long-term effects of hearing aid use on auditory spectral discrimination and temporal envelope sensitivity and speech perception in noise. The Journal of International Advanced Otology, 18(1), 43. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Anderson ES, Nelson DA, Kreft H, Nelson PB, & Oxenham AJ (2011). Comparing spatial tuning curves, spectral ripple resolution, and speech perception in cochlear implant users. The Journal of the Acoustical Society of America, 130(1), 364–375. 10.1121/1.3589255 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Anderson ES, Oxenham AJ, Nelson PB, & Nelson DA (2012). Assessing the role of spectral and intensity cues in spectral ripple detection and discrimination in cochlear-implant users. The Journal of the Acoustical Society of America, 132(6), 3925–3934. 10.1121/1.4763999 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Aronod JM, & Landsberger DM (2013). The development of a modified spectral ripple test. The Journal of the Acoustical Society of America, 134(2), EL217–EL222. 10.1121/1.4813802 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Attaheri A, Choisdealbha ÁN, Di Liberto GM, Rocha S, Brusini P, Mead N, Olawole-Scott H, Boutris P, Gibbon S, Williams I, Grey C, Flanagan S, & Goswami U. (2022). Delta- and theta-band cortical tracking and phase-amplitude coupling to sung speech by infants. NeuroImage, 247, 118698. 10.1016/j.neuroimage.2021.118698 [DOI] [PubMed] [Google Scholar]
  6. Biot L, Jacxsens L, Cardon E, Versnel H, Rhebergen KS, Boerboom RA, Gilles A, Van Rompaey V, & Lammers MJW (2024). Validation of the acoustic change complex (ACC) prediction model to predict speech perception in noise in adult patients with hearing loss: A study protocol. Diagnostic and Prognostic Research, 8(1), 1. 10.1186/s41512-024-00164-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Brennan MA, & McCreery RW (2021). Audibility and spectral-ripple discrimination thresholds as predictors of word recognition with nonlinear frequency compression. Journal of the American Academy of Audiology, 32(9), 596–605. 10.1055/s-0041-1732333 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Brown CJ, Etler C, He S, O’Brien S, Erenberg S, Kim J-R, Dhuldhoya AN, & Abbas PJ (2008). The electrically evoked auditory change complex: Preliminary results from nucleus cochlear implant users. Ear and Hearing, 29(5), 704. 10.1097/AUD.0b013e31817a98af [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Byrne D, Dillon H, Tran K, Arlinger S, Wilbraham K, Cox R, Hagerman B, Hetu R, Kei J, Lui C, Kiessling J, Kotby MN, Nasser NHA, El Kholy WAH, Nakanishi Y, Oyer H, Powell R, Stephens D, Meredith R, … Ludvigsen C. (1994). An international comparison of long-term average speech spectra. The Journal of the Acoustical Society of America, 96(4), 2108–2120. 10.1121/1.410152 [DOI] [Google Scholar]
  10. Cheek D, & Cone B. (2020). Evidence of vowel discrimination provided by the acoustic change complex. Ear and Hearing, 41(4), 855. 10.1097/AUD.0000000000000809 [DOI] [PubMed] [Google Scholar]
  11. Chen KH, & Small SA (2015). Elicitation of the acoustic change complex to long-duration speech stimuli in four-month-old infants. International Journal of Otolaryngology, 2015, 1–12. 10.1155/2015/562030 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Ching TYC, Dillon H, Button L, Seeto M, Van Buynder P, Marnane V, Cupples L, & Leigh G. (2017). Age at intervention for permanent hearing loss and 5-year language outcomes. Pediatrics, 140(3). 10.1542/peds.2016-4274 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Ching TYC, Zhang VW, Ibrahim R, Bardy F, Rance G, Van Dun B, Sharma M, Chisari D, & Dillon H. (2023). Acoustic change complex for assessing speech discrimination in normal-hearing and hearing-impaired infants. Clinical Neurophysiology, 149, 121–132. 10.1016/j.clinph.2023.02.172 [DOI] [PubMed] [Google Scholar]
  14. Cone BK, Smith S, & Cheek Smith DE (2022). Acoustic change complex and visually reinforced infant speech discrimination measures of vowel contrast detection. Ear and Hearing, 43(2), 531–544. 10.1097/AUD.0000000000001116 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Cone-Wesson B. (2003). Pediatric audiology: A review of assessment methods for infants. Hearing Balance and Communication, 1(3), 175. 10.1080/16513860310001979 [DOI] [Google Scholar]
  16. Culbertson SR, Dillon MT, Richter ME, Brown KD, Anderson MR, Hancock SL, & Park LR (2022). Younger age at cochlear implant activation results in improved auditory skill development for children with congenital deafness. Journal of Speech, Language, and Hearing Research: JSLHR, 65(9), 3539–3547. 10.1044/2022_JSLHR-22-00039 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Delorme A, & Makeig S. (2004). EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. Journal of Neuroscience Methods, 134(1), 9–21. 10.1016/j.jneumeth.2003.10.009 [DOI] [PubMed] [Google Scholar]
  18. Dimitrijevic A, Michalewski HJ, Zeng FG, Pratt H, & Starr A. (2008). Frequency changes in a continuous tone: Auditory cortical potentials. Clinical Neurophysiology, 119(9), 2111–2124. 10.1016/j.clinph.2008.06.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. DiNino M, & Arenberg JG (2018). Age-related performance on vowel identification and the spectral-temporally modulated ripple test in children with normal hearing and with cochlear implants. Trends in Hearing, 22, 2331216518770959. 10.1177/2331216518770959 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Eisenberg LS, Johnson KC, Martinez AS, Cokely CG, Tobey EA, Quittner AL, Fink NE, Wang N-Y, Niparko JK, & CDaCI Investgative Team. (2006). Speech recognition at 1-year follow-up in the childhood development after cochlear implantation study: Methods and preliminary findings. Audiology & Neuro-Otology, 11(4), 259–268. 10.1159/000093302 [DOI] [PubMed] [Google Scholar]
  21. Eisenberg LS, Martinez AS, & Boothroyd A. (2007). Assessing auditory capabilities in young children. International Journal of Pediatric Otorhinolaryngology, 71(9), 1339–1350. 10.1016/j.ijporl.2007.05.017 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Friesen LM, & Tremblay KL (2006). Acoustic change complexes recorded in adult cochlear implant listeners: Ear and Hearing, 27(6), 678–685. 10.1097/01.aud.0000240620.63453.c3 [DOI] [PubMed] [Google Scholar]
  23. Han JH, & Dimitrijevic A. (2015). Acoustic change responses to amplitude modulation: A method to quantify cortical temporal processing and hemispheric asymmetry. Frontiers in Neuroscience, 9. 10.3389/fnins.2015.00038 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. He S, Grose JH, & Buchman CA (2012). Auditory discrimination: The relationship between psychophysical and electrophysiological measures. International Journal of Audiology, 51(10), 771–782. 10.3109/14992027.2012.699198 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Henry BA, Turner CW, & Behrens A. (2005). Spectral peak resolution and speech recognition in quiet: Normal hearing, hearing impaired, and cochlear implant listeners. The Journal of the Acoustical Society of America, 118(2), 1111–1121. 10.1121/1.1944567 [DOI] [PubMed] [Google Scholar]
  26. Horn D, Dudley DJ, Dedhia K, Nie K, Drennan WR, Won JH, Rubinstein JT, & Werner LA (2017). Effects of age and hearing mechanism on spectral resolution in normal hearing and cochlear-implanted listeners. The Journal of the Acoustical Society of America, 141(1), 613–623. 10.1121/1.4974203 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Horn D, Won JH, Rubinstein JT, & Werner LA (2017). Spectral ripple discrimination in normal hearing infants. Ear and Hearing, 38(2), 212–222. 10.1097/AUD.0000000000000373 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Joint Committee on Infant Hearing. (2019). Year 2019 position statement: Principles and guidelines for early hearing detection and intervention programs. The Journal of Early Hearing Detection and Intervention, 4(2), 1–44. [Google Scholar]
  29. Kalaiah MK (2018). Acoustic change complex for frequency changes. Hearing, Balance and Communication, 16(1), 29–35. 10.1080/21695717.2017.1421813 [DOI] [Google Scholar]
  30. Kalashnikova M, Peter V, Di Liberto GM, Lalor EC, & Burnham D. (2018). Infant-directed speech facilitates seven-month-old infants’ cortical tracking of speech. Scientific Reports, 8(1), 13745. 10.1038/s41598-018-32150-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Kim J-R (2015). Acoustic Change Complex: Clinical Implications. Journal of Audiology and Otology, 19(3), 120–124. 10.7874/jao.2015.19.3.120 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Korver AMH, Konings S, Dekker FW, Beers M, Wever CC, Frijns JHM, Oudesluys-Murphy AM, & DECIBEL Collaborative Study Group, for the. (2010). Newborn hearing screening vs later hearing screening and developmental outcomes in children with permanent childhood hearing impairment. JAMA, 304(15), 1701–1708. 10.1001/jama.2010.1501 [DOI] [PubMed] [Google Scholar]
  33. Lau BK, Oxenham AJ, & Werner LA (2021). Infant pitch and timbre discrimination in the presence of variation in the other dimension. Journal of the Association for Research in Otolaryngology, 22(6), 693–702. 10.1007/s10162-021-00807-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Liang C, Houston LM, Samy RN, Abedelrehim LMI, & Zhang F. (2018). Cortical processing of frequency changes reflected by the acoustic change complex in adult cochlear implant users. Audiology and Neurotology, 23(3), 152–164. 10.1159/000492170 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Martin BA, & Boothroyd A. (1999). Cortical, auditory, event-related potentials in response to periodic and aperiodic stimuli with the same spectral envelope. Ear and Hearing, 20(1), 33–44. [DOI] [PubMed] [Google Scholar]
  36. Martin BA, Boothroyd A, Ali D, & Leach-Berth T. (2010). Stimulus presentation strategies for eliciting the acoustic change complex: Increasing efficiency. Ear and Hearing, 31(3), 356. 10.1097/AUD.0b013e3181ce6355 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Martinez A, Eisenberg L, Boothroyd A, & Visser-Dumont L. (2008). Assessing speech pattern contrast perception in infants: Early results on VRASPAC. Otology & Neurotology: Official Publication of the American Otological Society, American Neurotology Society [and] European Academy of Otology and Neurotology, 29(2), 183–188. 10.1097/MAO.0b013e3181625114 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Martinez AS, Eisenberg LS, & Boothroyd A. (2013). The acoustic change complex in young children with hearing loss: A preliminary study. Seminars in Hearing, 34(4), 278–287. 10.1055/s-0033-1356640 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Moeller MP, Tomblin JB, Yoshinaga-Itano C, Connor CM, & Jerger S. (2007). Current state of knowledge: Language and literacy of children with hearing impairment. Ear and Hearing, 28(6), 740–753. 10.1097/AUD.0b013e318157f07f [DOI] [PubMed] [Google Scholar]
  40. Näätänen R, Paavilainen P, Rinne T, & Alho K. (2007). The mismatch negativity (MMN) in basic research of central auditory processing: A review. Clinical Neurophysiology, 118(12), 2544–2590. 10.1016/j.clinph.2007.04.026 [DOI] [PubMed] [Google Scholar]
  41. Nechaev DI, Milekhina ON, Tomozova MS, & Supin A. Ya. (2024). Discrimination of rippled spectra with different ripple widths in listeners with normal and impaired hearing. Neuroscience and Behavioral Physiology, 54(7), 1035–1041. 10.1007/s11055-024-01707-6 [DOI] [Google Scholar]
  42. Niparko JK, Tobey EA, Thal DJ, Eisenberg LS, Wang NY, Quittner AL, Fink NE, & CDaCI Investigative Team. (2010). Spoken language development in children following cochlear implantation. JAMA, 303(15), 1498–1506. 10.1001/jama.2010.451 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Noble AR, Halverson DM, Resnick J, Broncheau M, Rubinstein JT, & Horn DL (2024). Spectral resolution and speech perception in cochlear implanted school-aged children. Otolaryngology–Head and Neck Surgery, 170(1), 230–238. 10.1002/ohn.408 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Noble AR, Resnick J, Broncheau M, Klotz S, Rubinstein JT, Werner LA, & Horn DL (2023). Spectrotemporal modulation discrimination in infants with normal hearing. Ear and Hearing, 44(1), 109. 10.1097/AUD.0000000000001277 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Ostrod JM, Martin BA, & Boothroyd A. (1998). Cortical evoked response to acoustic change within a syllable. Ear and Hearing, 19(4), 290. [DOI] [PubMed] [Google Scholar]
  46. Small SA, & Werker JF (2012). Does the acc have potential as an index of early speech-discrimination ability? A preliminary study in 4-month-old infants with normal hearing: Ear and Hearing, 33(6), e59–e69. 10.1097/AUD.0b013e31825f29be [DOI] [PubMed] [Google Scholar]
  47. Strahm S, Small SA, Chan S, Tian DY, & Sharma M. (2022). The maturation of the acoustic change complex in response to iterated ripple noise in ‘normal’-hearing infants, toddlers, and adults. Journal of the American Academy of Audiology, 33, 301–310. 10.1055/a-1862-0198 [DOI] [PubMed] [Google Scholar]
  48. Supin A. Ya., Popov VV, Milekhina ON, & Tarakanov MB (1999). Ripple depth and density resolution of rippled noise. The Journal of the Acoustical Society of America, 106(5), 2800–2804. 10.1121/1.428105 [DOI] [PubMed] [Google Scholar]
  49. Tomblin JB, Harrison M, Ambrose SE, Walker EA, Oleson JJ, & Moeller MP (2015). Language Outcomes in Young Children with Mild to Severe Hearing Loss: Ear and Hearing, 36, 76S–91S. 10.1097/AUD.0000000000000219 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Tremblay KL, Billings CJ, Friesen LM, & Souza PE (2006). Neural representation of amplified speech sounds. Ear and Hearing, 27(2), 93. 10.1097/01.aud.0000202288.21315.bd [DOI] [PubMed] [Google Scholar]
  51. Tremblay KL, Kalstein L, Billings CJ, & Souza PE (2006). The neural representation of consonant-vowel transitions in adults who wear hearing aids. Trends in Amplification, 10(3), 155–162. 10.1177/1084713806292655 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Uhler Hunter, K. S, Tierney E, & Gilley PM (2018). The relationship between mismatch response and the acoustic change complex in normal hearing infants. Clinical Neurophysiology, 129(6), 1148–1160. 10.1016/j.clinph.2018.02.132 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Uhler K, Yoshinaga-Itano C, Gabbard SA, Rothpletz AM, & Jenkins H. (2011). Longitudinal infant speech perception in young cochlear implant users. Journal of the American Academy of Audiology, 22(3), 129–142. 10.3766/jaaa.22.3.2 [DOI] [PubMed] [Google Scholar]
  54. Uhler Warner-Czyz, A., Gidord R, & Working Group, P. (2017). Pediatric minimum speech test battery. Journal of the American Academy of Audiology, 28(3), 232–247. 10.3766/jaaa.15123 [DOI] [PubMed] [Google Scholar]
  55. Undurraga JA, Van Yper L, Bance M, McAlpine D, & Vickers D. (2021). Neural encoding of spectro-temporal cues at slow and near speech-rate in cochlear implant users. Hearing Research, 403, 108160. 10.1016/j.heares.2020.108160 [DOI] [PubMed] [Google Scholar]
  56. Winn MB, & O’Brien G. (2022). Distortion of spectral ripples through cochlear implants has major implications for interpreting performance scores. Ear and Hearing, 43(3), 764. 10.1097/AUD.0000000000001162 [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Won JH, Clinard CG, Kwon S, Dasika VK, Nie K, Drennan WR, Tremblay KL, & Rubinstein JT (2011). Relationship between behavioral and physiological spectral-ripple discrimination. Journal of the Association for Research in Otolaryngology, 12(3), 375–393. 10.1007/s10162-011-0257-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Won JH, Drennan WR, & Rubinstein JT (2007). Spectral-ripple resolution correlates with speech reception in noise in cochlear implant users. Journal of the Association for Research in Otolaryngology, 8(3), 384–392. 10.1007/s10162-007-0085-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Yoshinaga-Itano C, Sedey AL, Coulter DK, & Mehl AL (1998). Language of early- and later-identified children with hearing loss. Pediatrics, 102(5), 1161–1171. 10.1542/peds.102.5.1161 [DOI] [PubMed] [Google Scholar]
  60. Yoshinaga-Itano C, Sedey AL, Wiggin M, & Mason CA (2018). Language outcomes improved through early hearing detection and earlier cochlear implantation. Otology & Neurotology, 39(10), 1256–1263. 10.1097/MAO.0000000000001976 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The datasets generated during and/or analyzed during the current study are available from the corresponding author upon request.

RESOURCES