A systematic review of acoustic change complex (ACC) measurements and applicability in children for the assessment of the neural capacity for sound and speech discrimination

Objective: Theacousticchangecomplex(ACC)isacorticalauditoryevokedpotential(CAEP)andcanbeelicitedbya change in an otherwise continuous sound. The ACC has been highlighted as a promising tool in the assessment of sound and speech discrimination capacity, and particularly for difficult-to-test populations such as infants with hearing loss, due to the objective nature of ACC measurements. Indeed, there is a pressing need to develop further meanstoaccuratelyandthoroughlyestablishthehearingstatusofchildrenwithhearingloss,tohelpguidehearing interventions in a timely manner. Despite the potential of the ACC method, ACC measurements remain relatively rare in a standard clinical settings. The objective of this study was to perform an up-to-date systematic review on ACC measurements in children, to provide greater clarity and consensus on the possible methodologies, applications, and performance of this technique, and to facilitate its uptake in relevant clinical settings. Design: Original peer-reviewed articles conducting ACC measurements in children ( < 18 years). Data were extracted and summarised for: (1) participant characteristics; (2) ACC methods and auditory stimuli; (3) information related to the performance of the ACC technique; (4) ACC measurement outcomes, advantages, and challenges. The systematic review was conducted using PRISMA guidelines for reporting and the methodological quality of included articles was assessed. Results: A total of 28 studies were identified (9 infant studies). Review results show that ACC responses can be measured in infants (from < 3 months), and there is evidence of age-dependency, including increased robustness of the ACC response with increasing childhood age. Clinical applications include the measurement of the neural capacity for speech and non-speech sound discrimination in children with hearing loss, auditory neuropathy spectrum disorder (ANSD) and central auditory processing disorder (CAPD). Additionally, ACCs can be recorded in children with hearing aids, auditory brainstem implants, and cochlear implants, and ACC results may guide hearing intervention/rehabilitation strategies. The review identified that the time taken to perform ACC measurements was often lengthy; the development of more efficient ACC test procedures for children would be beneficial. Comparisons between objective ACC measurements and behavioural measures of sound discrimination showed significant correlations for some, but not all, included studies. Conclusions: ACC measurements of the neural capacity to discriminate between speech and non-speech sounds are feasible in infants and children, and a wide range of possible clinical applications exist, although more time-efficient procedures would be advantageous for clinical uptake. A consideration of age and maturational effects is recommended, and further research is required to investigate the relationship between objective ACC measures and behavioural measures of sound and speech perception for effective clinical implementation.


Introduction
Historically, the study of audiology has often focused on the outer, middle, and inner ear and the ability of the hearing system to detect sounds.The functions of the peripheral hearing system are now well understood (for example, Moore, 2012;W.A. Yost, 2013).There is significant interest, however, in the ability to not only detect sounds, but also to discriminate between dynamically variable sounds and the neural capacity available to perform this crucial task, since the ability to discriminate between changing sounds is vital to understanding the acoustically variable pattern of speech sounds (W.A. Yost, 2013).A desire to better understand acoustic discrimination has fueled interest in the exploration of pathways higher up the auditory system.An emerging technique, namely the 'Acoustic Change Complex' (ACC), is a promising candidate to probe the neural capacity to discriminate between changing sounds (Cheek and Cone, 2020;Kim, 2015;Martin et al., 2008).
The ACC is a cortical auditory evoked potential (CAEP) (Picton, 2010); this neural response originates in the auditory cortex (Martin et al. 2008;Picton 2010).An ACC is evoked by changes in an otherwise steady state auditory stimulus (Martin, et al., 2008) and is recorded by placing surface electrodes on the scalp measuring small varying voltages (Ostroff, et al., 1998).First, the onset of a sound stimulus yields a simple CAEP waveform 'onset response' (Martin and Boothroyd, 1999), corresponding to the detection of a sound (Visram, et al., 2023).When an auditory change is introduced (e.g. by disrupting the frequency, temporal properties or intensity of an acoustic stimulus), then multiple waveforms are generated, not only in response to the stimulus onset, but also in response to the stimulus change(s) and stimulus offset (Ostroff et al. 1998;Martin et al. 2008).Hence, these resulting waveforms, trailing after the onset response, are termed the ACC (Martin andBoothroyd 1999, 2000;Martin et al. 2008), and thus probe sound discrimination, as opposed to pure detection alone.The morphology of an ACC waveform is observed by plotting amplitude (µV) against time/latency (ms); the mature adult waveform consists of a positive peak (P1), then a negative peak (N1), followed by another positive peak (P2) (Martin et al. 2008;Picton 2010).
Interestingly, ACC measurements show promise clinically; ACCs display favourable test-retest reliability (Tremblay, et al., 2003), and good signal to noise ratios can be obtained with few stimulus presentations (Kim, 2015;Martin, et al., 2008).Further, ACCs have been observed following small amplitude changes (e.g. a couple of decibels), that are comparable to the just detectable psychoacoustic limits of the auditory system (Martin and Boothroyd, 2000).Additionally, ACC measurements of speech discrimination capacity in adults (e.g.vowel contrast detection) show high levels of sensitivity (up to 97%) with respect to perceptual measures of adult speech discrimination (Cheek and Cone, 2020).
ACC measurements have another inherent advantage.Since the testing method does not require a behavioural response from a subject (Kim, 2015;Martin, et al., 2008), the results provide an unbiased measurement of the capacity to perform sound discrimination.Therefore, ACC electrophysiological measurements have often been suggested as a (potentially) valuable tool for assessing auditory discrimination abilities of difficult-to-test populations, such as hearing impaired infants (Cone, et al., 2022;Cowan, et al., 2017;Small, 2015).Indeed, since the very first articles introducing the ACC, its potential for testing infants and children was noted (Martin and Boothroyd, 1999;Ostroff, et al., 1998).The prospect of a reliable objective test that can shed light onto speech discrimination capacity of children is like gold; infants (and some children) cannot easily be tested by behavioural methods (McCormick, 2004), still the ability to understand and discriminate between different speech sounds is crucial in the early years when language is developing (Lieu, et al., 2020;McCormick, 2004;Tomblin, et al., 2015;Yoshinaga-Itano, et al., 1998).Further, for hearing impaired infants, decisions regarding potential hearing interventions are time-sensitive, if to be maximally effective (e.g.cochlear implantation; Karltorp, et al., 2020;May-Mederake, 2012;Sharma, et al., 2020), and additional information regarding the auditory status could be valuable in guiding such interventional decisions.
Although the likely benefit of ACC measurements for children was first highlighted over twenty-five years ago (Ostroff, et al., 1998), ACC measurements in a standard clinical setting are (as of yet) relatively rare, and the maximum potential of this method is yet to be fully recognised (Kim, 2015;Sanju, et al., 2023).Over the years, a variety of studies have been published employing the ACC, and a wide range of methods and applications have been presented for this measurement (Kim, 2015).Although two other ACC reviews are available, only journal articles up until 2020 were considered for these earlier reviews, and some articles prior to 2020 were not included (Kim, 2015;Sanju, et al., 2023), and in recent years the number of publications has been on the rise.Hence, it would be advantageous to have an up-to-date systematic review on the ACC.Further, to the best of our knowledge, a systematic review focused on ACC measurements in children is not yet available.Given the potential of the ACC technique to investigate hearing impaired children, it seems advantageous to produce a systematic review specifically in relation to children's studies, with an aim to provide greater clarity and consensus on this tool, and to facilitate its uptake in clinical settings as relevant.Hence, the objective of this study is to perform a systematic review on ACC measurements in children, researching the methodologies, clinical applications, and outcomes.Further, we aim to develop a useful summary of the advantages and challenges associated with this technique as a guide for clinicians considering implementing ACC measurements on children in the future.

Protocol
The systematic review here was reported in compliance with the 'Preferred Reporting Items for Systematic Reviews and Meta-Analyses' (PRISMA) guidelines (Moher, et al., 2015;Page, et al., 2021).A protocol was designed, documented, and specified in advance according to the PRISMA guidelines; the guidelines consist of a 17-item checklist and details the rational and planned methods and analytical approach of the review (Supplemental Material 1).

Search strategy
A search strategy was created by a trained librarian, in communication with the authors.Studies were identified by searching electronic databases and scanning reference lists of included articles.Databases searched were Medline (Ovid), Embase, Web of Science Core Collection (Web of Knowledge), Cochrane Central Register of Controlled Trials (Wiley), CINAHL Plus (EBSCO), Scopus, and Google Scholar.The last search was performed on 15 February 2024 (see Supplemental Material 2 for the search strings used and a table showing the results).

Study selection
Eligibility assessment was performed independently in a standardised manner by two reviewers.A full description of eligibility criteria for the articles is provided in the Protocol (Supplemental Material 1).In summary, articles were included if they met the following eligibility criteria: (1) Participants: studies relating to children (< 18 years old); (2) Hearing Interventions: no restrictions (articles referencing hearing aids, cochlear implants (CI) and other implantable devices, and no devices, all included); (3) Methods: studies in which ACC measurements were performed on children were included (studies related to other similar CAEPs (e.g.mismatch negativity (MMN); Martin, et al., 2008) were only included if the articles also specifically conducted ACC measurements); electroencephalography (EEG) studies included, magnetoencephalography (MEG) studies excluded; (4) Outcomes: studies reporting outcomes related to the use of ACC measurements in children were included; (5) Study design: human studies included (animal studies excluded), systematic reviews, and meta analyses excluded; (6) Years considered: all years available online; (7) Language: articles written in English were included; (8) Publication status: published peer reviewed articles, available in full text, were included.
The Endnote reference software package was employed to manage records and data throughout the review.Two reviewers independently screened the titles and abstracts yielded by the search against the inclusion criteria.Full text articles were obtained for all titles that appeared to meet the eligibility criteria or where there was any uncertainty.Two reviewers then independently screened the full text articles and decided whether these met the eligibility criteria.Reviewers resolved disagreement via discussion and a third reviewer to adjudicate.Full texts of all eligible articles were read and assessed according to the eligibility criteria.

Study assessment
The quality of the included studies was assessed by 'The ACC Systematic Review Methodological Quality Assessment Checklist' (Supplemental Material 3); the checklist was developed in part from the Down's and Black Checklist for quality assessment of health care interventions (Downs and Black, 1998) and the Cochrane recommendations for systematic reviews (Higgins and Green, 2008).The checklist employed here (Supplemental Material 3) was customised in order to be relevant to the present systematic review, and with the aim to facilitate systematic analysis of the studies identified, whilst striving to make quality judgements in a more objective manner.Questions one to seven are relatively generic; questions eight to fourteen are more specifically related to ACC studies (Supplemental Material 3).Higher total scores indicate a greater methodological quality (maximum score of 14 points), and vice versa.

Data extraction
From all included studies, the following data were extracted: (1) Participant variables: number of participants, participant characteristics (age, hearing level, hearing interventions used (if any, e.g.hearing aids or CI, syndromes/diseases noted); (2) Intervention variables: the types of stimuli devised to evoke ACC responses, the equipment used, the ACC measurement methods and data analyses techniques employed, information relating to the time taken to perform the measurements; (3) Comparison variables: information relating to the performance of the ACC technique, e.g.comparisons between behavioural sound/speech discrimination tests and objective ACC measures; (4) Outcome variables: reported outcomes from ACC measurements, and advantages, and challenges, described to be associated with ACC measurements.There were no pre-planned data assumptions or simplifications.

Literature search and selection
The literature search resulted in 522 study reports.The search involved two phases.In the first phase of the literature search no restrictions were placed on participant ages (all studies on adults and children were included).In the second phase, only papers related to ACC children's studies were included (Fig. 1).
Phase (1): After the removal of duplicates (348 in total), the titles and abstracts of 174 articles were screened.This resulted in the exclusion of a total of 72 articles for not meeting the inclusion criteria.We excluded an additional 20 articles, based upon the full text assessment, and we included three articles through reference list examination.This resulted in the inclusion of 85 articles for review (Fig. 1).
Phase (2): From the 85 peer reviewed articles included in the first phase, 28 studies were identified that conducted ACC measurements on children (9 infant studies), after a total of 57 articles related to ACC measurements in adults were excluded.In summary, 28 articles relating to ACC measurements in children were included for review (Fig. 1).

Methodological quality assessment of included studies
Included articles were assessed by the ACC Systematic Review Methodological Quality Assessment Checklist (Supplemental Material 3).The average score for all the studies was good (mean=67%, SD=18), and most studies had reasonable scores (Supplemental Material 4).Scores for individual studies are provided for reference purposes if required (Supplemental Material 4).
The majority of studies were well written, for example, scoring highly on clearly describing the objective/hypothesis of the study (96% of studies), the patient characteristics (96% of studies), the inclusion/ exclusion criteria (86% of studies), the methods (79% of studies), and the main findings (96% of studies).Sometimes methodological details were missing, for example the sound level and the means of delivery of the evoking stimuli were not always reported.
Many, but not all, studies (71%) reported that ACC measurements were conducted in a sound attenuating booth.A little over half the articles (57%) reported participant factors that may influence ACC results, for example any history of neurological conditions.A high proportion of the papers took into account the participants' functional hearing abilities (79%), and in these cases hearing was tested by screening methods (infants), or pure tone audiometry (PTA) in older children, and sometimes tympanometry was also performed.Although most studies identified here described a procedure whereby the participant was given a quiet task during EEG recordings (e.g.watching a silent movie with captions), the participants' state (awake or asleep) was only reportedly actively monitored in 36% of studies.

Characteristics of study populations
Nine studies conducted ACC measurements on infants under 12 months of age (Table 1, Supplemental Material 5); the youngest successful reported age for ACC measurements was around two to three months of age (Ching, et al., 2023;Strahm, et al., 2022).Three studies conducted ACC measurements on one to two year old children, and six studies incorporated ACC measurements on two to five year olds; overall, a total of 13 studies conducted ACC measurements on zero to five year olds (Table 1, Supplemental Material 5).Further details of participant ages covered by the studies are shown in Table 1 and summarised in Supplemental Material 5.
The majority of the studies (20 articles) conducted ACC measurements on infants and children with hearing within normal limits, according to newborn hearing screens/PTA measurements (Table 1, Supplemental Material 6).Eight studies showed that the ACC can be a reliable tool in assessing hearing impaired children, including infants (Ching, et al., 2023), with hearing losses ranging from mild to profound (Table 1, Supplemental Material 6).Three studies focused on auditory neuropathy spectrum disorder (ANSD); ACC responses were advocated as a potential tool to investigate temporal resolution abilities in children with ANSD (He, et al., 2013(He, et al., , 2015;;McFayden, et al., 2020).Three studies employed ACC measurements to study sound discrimination in children with central auditory processing disorder (CAPD; Kumar, et al., 2020aKumar, et al., , 2021Kumar, et al., , 2020b)).One lone study employed auditory brainstem implants (ABI; He, et al., 2018).Cochlear implants (CI) were included in four studies (He, et al., 2014;Nassar, et al., 2022;Turkyilmaz, et al., 2013;Xie, et al., 2022).These studies demonstrate the utility of such measurements in investigating discrimination capacities in children fitted with a CI or ABI device (cf.Table 3 and Supplemental Material 6).Hearing aids were referenced in seven studies (Ching, et al., 2023;He, et al., 2015;Kosemihal and Akdas, 2021;Martinez, et al., 2013; S. Meehan et al. McFayden, et al., 2020;Shetty and Puttabasappa, 2020); it was shown that it is possible to record ACC responses from infants and children with hearing loss while wearing a hearing aid (Ching, et al., 2023;Martinez, et al., 2013;Shalaby, et al., 2022).Further details summarising the hearing levels and interventions are summarised in Table 1 and Supplemental Material 6.

ACC methodologies
In order to perform ACC measurements, the following are required: a suitable location (e.g. a sound attenuated booth), equipment to acquire an EEG recording (including an electrode montage), an acoustic or electrical stimulus to evoke the ACC response, an experimental procedure for the participant, and a data analysis approach.

Equipment and experimental set-up for recording ACCs in children
Equipment is required for the acquisition of EEG data; a list of the various kit employed by the included studies is summarised in Supplemental Material 7. The NeuroScan system designed by Compumedics (Charlotte, NC, USA) proved to be a popular brand, used in 16 of the studies, perhaps due to the multiple arrays of integrated hardware and software systems developed by this company for the acquisition, amplification, and analysis of EEG data.Three studies employed the Biologic Pro-Navigator auditory evoked potential (AEP) system (Natus Hearing, Middleton, WI, USA; Supplemental Material 7); this equipment has the potential to measure a range of other AEPs (e.g.including auditory brainstem response; ABR) and has accompanying software for data analysis.
Electrodes placed on the surface of the scalp are required to measure the ACC response (Martin, et al., 2008).The electrode placement location on the scalp was universally reported according to the well-renowned 'International 10-20 system' (Jasper, 1958).An array of electrode montages were reported (Supplemental Material 7).A number of studies used simpler set ups, cleaning a child's skin with abrasive gels (to reduce impedances) and applying just a few electrodes (He, et al., 2013;Small and Werker, 2012); at minimum, an active recording electrode is required, together with a reference electrode and a ground electrode (Supplemental Material 7).In other cases, more elaborate set ups were reported, using an electrode net/cap (Aghabeig, et al., 2019;Martin, et al., 2010;McCarthy, et al., 2019;Petley, et al., 2024;Turkyilmaz, et al., 2013), in which multiple channel EEG recordings could take place (Aghabeig, et al., 2019; Supplemental Material 7).

Acoustic and electrical stimuli used to evoke ACCs
The evoking stimuli can be classified as acoustic or electrical in origin.The acoustic stimuli consisted of short speech sounds, or nonspeech sounds in which the psychoacoustic properties were manipulated to enact an acoustic change event (Table 2).A total of 17 articles reported the use of speech sounds to evoke ACC responses in infants and children (Table 2).Example speech material, for instance incorporating intrinsic changes in frequency or temporal properties, included concatenated vowel tokens (Cone, et al., 2022), or consonant-vowel (CV) combinations (Chen and Small, 2015;Kumar, et al., 2021) (Table 2).
Intensity changes were also used to evoke an ACC response (Table 2).One study employed a range of intensity changes (up to 20 dB) mid-way through a 1 kHz tonal stimulus (total duration of 500 ms) (Kumar, et al., 2020b).Another study manipulated amplitude modulation rates (Petley, et al., 2024).In another study, ACC responses were elicited with combinations of white noise and iterated ripple noise (IRN; Strahm, et al., 2022).
Examples of the use of electrical stimuli to evoke an ACC response were also reported (He, et al., 2013(He, et al., , 2018(He, et al., , 2014;;Table 2).For example, in a study investigating CI users, an ACC was evoked by switching the delivery of biphasic pulse train stimuli between different CI electrodes     (He, et al., 2014).Many studies' outcomes surrounded the development of evoking stimuli effective in eliciting ACC responses in children, including the refining of both speech and non-speech stimuli so that ACC responses could be observed in the particular participant population (Table 3; El-Kholy, et al., 2023;Martin, et al., 2010).

The sound level and delivery method of the acoustic or electrical stimuli
The delivery method employed depended on the type of evoking stimulus being used, and whether it was acoustic or electrical (Supplemental Material 7).Acoustic stimuli were often delivered via a loudspeaker (ranging from 0.75 to 1.5 m from the participant, and usually at 0 • azimuth), and sometimes via insert earphones (Supplemental Material 7); naturally, appropriate calibration methods to ensure the correct sound level were generally reported (Cone, et al., 2022;Martin, et al., 2010).The sound level of the acoustic stimuli used varied between studies; in most studies, however, the level employed simply ensured that the sound was definitely audible, and sound levels of around 65 to 85 dB SPL were frequently reported (Table 2), roughly equivalent to the level of noisy speech.
Electrical stimuli were frequently employed for participants with a CI or ABI; the speech processor was bypassed, and electrical stimuli were delivered directly to individual ABI or CI electrodes (He, et al., 2013(He, et al., , 2014)).In CI studies, sound was delivered at the 'maximum comfortable level' (He, et al., 2013(He, et al., , 2018(He, et al., , 2014)).

Procedures for measuring ACCs in children
The procedures used during EEG acquisition were generally similar between articles and varied depending on the age of the participant (Supplemental Material 8).Infants were usually seated in a caregiver's lap, or a highchair, and distracted by an assistant with quiet ageappropriate toys or a silent child-friendly movie (Cone, et al., 2022;McCarthy, et al., 2019;Small and Werker, 2012).The assistant played a role in monitoring the child's arousal state and muscle movement and electrical artefact, so that testing could be paused if the infant started cry or vocalize continuously, move excessively or fall asleep (Chen and Small, 2015;Ching, et al., 2023;Cone, et al., 2022).Hence, breaks were reportedly taken where necessary, to help ensure the infant was quiet and alert (Chen and Small, 2015;Strahm, et al., 2022).
For younger children the caregiver and assistant often remained in the booth, as reported in one study testing from toddlers to six year old children (Martinez, et al., 2013).As children became older and able to sit unassisted, they could be seated in a comfortable chair and distracted by silent age-appropriate movies, potentially with closed captions if the child could read (He, et al., 2018;Kumar, et al., 2020b;Martin, et al., 2010;Shalaby, et al., 2022).Participants were instructed to remain awake, minimize movement (to reduce myogenic artefacts), and be as quiet as possible, and breaks were provided as needed to facilitate compliance with these instructions (He, et al., 2013;Kosemihal and Akdas, 2021;Turkyilmaz, et al., 2013).

Time taken to perform ACC measurements
Information related to the time taken to perform EEG measurements was extracted (if available) from the included articles and is summarised in Table 2.The average total time for performing EEG measurements (including preparation time/set up/electrode application) was 1.7 h, and generally testing was reported to be completed in a single session.Even children under two years of age reportedly managed testing sessions of a couple of hours (He, et al., 2015).

Data analysis techniques in included studies
A number of steps are involved in preparing EEG data for analysis following its acquisition, including digitization, amplification, filtering, and artefact rejection (Chen and Small, 2015;Martin, et al., 2010;Martinez, et al., 2013).The filtering step is required to remove interference from electromagnetic interference, or muscle contractions, for example band-pass filtering to remove below around 0.1 Hz and above 30 Hz (Martin, et al., 2010;Petley, et al., 2024).
The ACC waveforms tend to be quite noisy, and artefact rejection is required, for example to remove artefacts related to myogenic noise and amplitude spikes related to eye blinks (Kumar, et al., 2021;Martinez, et al., 2013;Shetty and Puttabasappa, 2020).Artefact rejection methods included more traditional approaches (at least 14 studies), such as rules-based visual detection and threshold values (e.g. if a peak is above/below a certain threshold it is eliminated; Chen and Small, 2015;Cone, et al., 2022;He, et al., 2013He, et al., , 2015He, et al., , 2018)), or applying artefact rejection algorithms (at least five studies; e.g.Ching, et al., 2023;Martinez, et al., 2013).Occasionally, individual component analysis (ICA) techniques were applied (two studies; Aghabeig, et al., 2019;Petley, et al., 2024); ICA is a signal processing strategy that can be used to separate multivariate signals into additive components, and can be used to remove ocular, muscular, and cardiac artefacts from EEG signals (Uriguen and Garcia-Zapirain, 2015).One study employed tensor decomposition methods; this technique enables, not only the investigation of temporal, but also spatial and frequency characteristics simultaneously, providing a multi-domain technique suitable for complex ACC data (Aghabeig, et al., 2019).
Once artefact rejection has been performed, epochs (or 'time windows') are selected from continuous EEG data for analysis based on the time related to the onset of the evoking stimulus (Martinez, et al., 2013), for instance selecting an epoch 900 ms in length (− 100 ms prestimulus to 800 ms posttimulus; Chen and Small, 2015).Or, as another example, a total recording time of 1500 ms, including 100 ms prestimulus time was selected (Strahm, et al., 2022).
Occasionally, the CAEP responses were assessed independently by     two researchers blind to subject identification and stimulus conditions (He, et al., 2014) to reduce potential bias from the more qualitative approaches.

Comparison variables and the performance of the ACC technique
First, the performance of ACC measurements relative to psychophysical discrimination thresholds will be discussed (four studies; Table 3), and secondly reports describing the performance of ACC responses compared to speech discrimination will be addressed (five studies; Table 3).
First, significant correlations were identified between behavioural psychophysical measures and objective ACC responses (in two studies).More specifically, a moderate correlation was found with behavioural difference limen for intensity measures (DLIs; Kumar, Singh, et al. 2020), and a strong correlation was reported with auditory fusion test (AFT) results (Nassar et al. 2022).By contrast, ACC measures were not found to correlate significantly with amplitude modulation (AM) behavioural depth thresholds, in a study investigating children with listening difficulties (Petley et al. 2024).Also, in a study comparing objective ACC gap detection thresholds (GDTs) to behavioural GDTs (for ABI children), both objective and behavioural GDTs showed inter-and intra-subject variations, and consistency between the measures was not always observed (He et al. 2018).
Secondly, five studies compared ACC responses to speech

(Small 2015)
This study stated that ACC responses were recorded for infants in response to CVCV (consonant-vowelconsonant-vowel) speech stimuli, (however, limited data was presented in the paper to support the claim).
n/a 24 (Small and Werker 2012) This work demonstrated that ACC responses could be recorded in infants (4 months old) in response to synthetic consonant pairs.The authors hence highlighted the potential of the ACC measurements to be used as an index of discrimination for infants.
n/a (continued on next page) S. Meehan et al. discrimination.For example, a comparison of ACC measurements to behavioural responses obtained via a visually reinforced infant speech discrimination (VRISD) paradigm (assessing vowel discrimination), revealed a superiority in the ACC responses (Cone et al. 2022).The ACC responses were more robust; the sensitivity of ACC responses was over 90%, whereas for VRISD it was around 70% (Cone et al. 2022).Three further studies identified significant moderate to strong correlations between ACC measures and speech discrimination, with a variety of different types of speech material being employed including words and sentences in both quiet and noise (He, et al., 2013(He, et al., , 2015;;Xie, et al., 2022).In a fifth study, no correlations were identified between ACC responses (evoked by speech in noise stimuli) and behavioural speech in noise tests using standardised Arabic word/sentences in noise tests in children (Shalaby et al. 2022).
In a final example, objective ACC measurements of speech sound discrimination correlated weakly with parent-rated auditory functional performance in a study testing infants with hearing aids (Ching et al. 2023).
Some studies considered how the ACC responses compared to the level of hearing loss (Table 3).For example, ACC P1 and N1 latencies were significantly longer for a sensorineural hearing loss (SNHL) group relative to normal hearing controls (Hamdy, et al., 2024;Table 3).A second study similarly found a significant difference between the ACC latencies in normal hearing versus hearing-impaired groups (Kosemihal and Akdas, 2021; Table 3).Further, the ACC responses of normal hearing infants were significantly higher than those with severe hearing loss in an additional study (Ching, et al., 2023; Table 3).Generally, however, ACC methods were designed to investigate the neural capacity for sound discrimination, rather than to differentiate between different levels of hearing loss, and so testing was usually conducted at supra-threshold levels (cf.Table 2).The onset CAEP response is better suited to detecting the degree of hearing loss, since this measure corresponds to sound detection rather than discrimination, and onset CAEP responses can be acquired in a relatively short time (e.g.24 min; Visram, et al., 2023).

ACC measurements in children; advantages and challenges
The outcomes for the individual studies are summarised in Table 3, and the advantages and challenges surrounding ACC measurements in children are listed below.
Advantages included: (1) The ACC can be recorded in the absence of attention and is a useful objective index of the neural capacity for auditory discrimination in children, including infants and toddlers (Chen and Small, 2015;Cone, et al., 2022;Martin, et al., 2010;Small, 2015;Small and Werker, 2012;Strahm, et al., 2022).This contrasts to traditional methods of sound discrimination (e.g.VRISD), which require a behavioural response and often the training of infants to respond to speech sounds (Cone, et al., 2022).(2) Excellent test-retest reliability was shown at the individual participant level in children (Martin, et al., 2010).
(3) ACC measurements can sometimes be more sensitive, and more reliable, than a behavioural measure of speech sound discrimination in infants (Cone, et al., 2022).(4) ACC measurements can be employed for children with hearing loss (Ching, et al., 2023;Martinez, et al., 2013;Shalaby, et al., 2022), CAPD, and ANSD to explore sound discrimination abilities, including temporal resolution deficits (He, et al., 2013;Kumar, et al., 2020b).( 5) ACCs can be recorded in young children and infants (as young as three months old), both with and without hearing aids, and can be used to investigate sound discrimination capacities for child hearing aid users (Ching, et al., 2023;Martinez, et al., 2013;Shalaby, et al., 2022).( 6) ACC measurements may potentially be used to fine tune hearing aids and evaluate hearing aid performance (Cone, et al., 2022;Kosemihal and Akdas, 2021;Shalaby, et al., 2022;Shetty and Puttabasappa, 2020;Xie, et al., 2022).( 7) The ACC technique can be applied to children with CI and ABI (cf. Table 1, Table 3).( 8) ACC measurements may be able to help steer hearing intervention choices from a young age and help objectively identify CI candidates (He, et al., 2015;Martinez, et al., 2013), along with guiding rehabilitation strategies and auditory training programmes for children with hearing difficulties (He, et al., 2013;Nassar, et al., 2022;Petley, et al., 2024).
Challenges included: (1) Maturational effects and the variability of ACC responses with age can complicate ACC measurements; age effects were reported in a number of studies (Ching, et al., 2023;Hamdy, et al., 2024;Petley, et al., 2024;Strahm, et al., 2022; Table 3).Age effects may also present challenges for the derivation of optimal stimuli parameters (Small and Werker, 2012;Strahm, et al., 2022).(2) There is a higher neural refractoriness in children compared to adults, presenting a challenge for ACC acquisition; slower rates of stimulus presentations may be required to give the neurons chance to recover between stimuli presentations (Chen and Small, 2015;Martin, et al., 2010).(3) Data analysis for ACC measurements can be complex and challenging, for instance due to overlapping waveforms, or a large number of artefacts arising related to movement, particularly in mobile young children (Aghabeig, et al., 2019;Cone, et al., 2022;Turkyilmaz, et al., 2013).For example, one particular study found that noise in the averaged waveforms was higher for the children compared to the adults (Martin, et al., 2010).(4) A number of included articles highlighted that measuring the ACC can be time-consuming (He, et al., 2015;Martin, et al., 2010;Martinez, et al., 2013), since ACC acquisition requires the subject to be alert, quiet, and sit still long enough to place the electrodes, measure impedances, and deliver the stimuli.(5) Hearing aids may contaminate ACC results (by affecting ACC response traces), and the contribution of hearing aid signal processing on evoked neural activity is yet to be fully defined (Hamdy, et al., 2024;Martinez, et al., 2013).

Discussion
With this systematic review we aimed to investigate the potential of ACC measurements in children.
Given the results from the methodological quality review of the 28 included articles (Supplemental Material 3 & 4), the following is recommended in future studies conducting ACC measurements in children: (1) taking participants functional hearing abilities into account; (2) a consideration of participant factors that may influence ACC results (e.g. a history of neurological conditions); (3) a complete detailed description of ACC methods; (4) the use of a sound attenuated booth for ACC measurements; (5) direct monitoring of the participant level of attention during ACC recordings; and (6) a consideration of the possibility of blinding individuals analysing ACC data.

The feasibility of ACC measurements in children
Given all the included articles produced reports of successful recordings of ACC responses, clearly ACC measurements in both infants and children are entirely feasible, even for very young subjects (e.g. for infants around two to three months of age; Ching, et al., 2023;Strahm, et al., 2022).The choice of evoking stimulus can, however, affect the reliability of the ACC response, as observed in one study using various speech sounds (different types of CV combinations), to elicit ACC responses in four month old infants (Small and Werker, 2012).
The reliability of the ACC response is also age-dependent; the robustness of the ACC response may improve with age, since peak latencies and amplitudes varied with age and maturational effects were observed (Hamdy, et al., 2024;Martin, et al., 2010;Petley, et al., 2024;Strahm, et al., 2022).For example, in a study measuring ACCs in infants, toddlers, and adults, the ACC responses were noticeably more reliable with increasing age, so that by adulthood ACC responses were readily detectable (Strahm, et al., 2022).In this study, the ACC waveform morphology varied with age; the P1 peak was observed in infancy, however the N1 peak only started to become visible from the toddler age group onwards (Strahm, et al., 2022).By adulthood, the complete P1-N1-P2 waveform was observed (Strahm, et al., 2022).Another study also showed that young children gave rise to P1-N2 ACC responses, and the N1 peak was similarly not observed (Martinez, et al., 2013).A further study observed that the ACC N1 component showed a correlation with age; the older children presented with larger N1 responses (i.e. more negative; Petley, et al., 2024).It is perhaps to be expected that the ACC waveforms vary with age, since the morphological characteristics of CAEP waveforms are known to vary with age (Wunderlich and Cone-Wesson, 2006).For example, maturational effects on CAEP morphology have been described previously; in children under five years of age the CAEP waveform is dominated by a P1 peak and the N1 peak is not observable; the N1 peak becomes more visible from around eight years old (Gilley, et al., 2005).
Another ACC study reported a trend towards shorter latencies of N1 and P2 with increasing age and found that P1 amplitudes were significantly larger in children than adults (Hamdy, et al., 2024); it is notable, however, that the youngest participants in this study were already ten years of age and the adults ranged up to 50 years old.By comparison, the latency of the P1 CAEP 'onset' response is known to decrease rapidly over the first two to three years of life when neural plasticity is at its highest, then more gradually, reducing from values of around 300 ms for newborns through to values of around 60 ms for adults (Sharma, et al., 2015).It is intuitive that as CAEPs (both the onset and ACC response) become increasingly mature with age that the neurons are able to respond more rapidly/efficiently and recover faster between different stimuli, and hence latencies decrease, and responses are more readily observed.Correspondingly, one of the included studies found that by increasing stimulus length and lengthening the time between stimuli presentations, allowing more time to accommodate the longer neuronal refractory period, better-defined ACC waveforms were observed for infants (Chen and Small, 2015).
Overall, results further highlight the need to derive age-specific normative data, and age-appropriate stimulus types and parameters.A more detailed investigation into effects of maturation on the ACC response would be beneficial in future, to determine the reliability of the ages from which ACC response levels become sufficiently mature to observe and to facilitate the acquisition of ACC responses in the very youngest of children.It is also important to be aware of these age effects when analysing and interpreting ACC data for all children.

Methods for conducting ACC measurements in children
It is interesting to consider the methods employed to evoke and record ACCs and how this relates to the patient experience.On the positive side, the experience for a participant may be relatively pleasant, since the majority of acquisition time involved sitting in a comfortable chair and watching closed caption movies, or (if younger) sitting in a caregiver's lap and being entertained by an assistant with ageappropriate toys (Supplemental Material 8).On the downside, ACC acquisition could often be a lengthy process, taking around a couple of hours (Table 2), and it was often reported that ensuring participants were still, alert, and quiet during this time could be challenging, particularly for the very youngest subjects (Supplemental Material 8).Further, although the possibility of letting subjects sleep during ACC data acquisition may be tempting, it is presently non-desirable, since sleep can hinder ACC acquisition (Lightfoot, 2016;Uhler, et al., 2018).The fact that ACC measurements were not recordable in sleeping infants it is not surprising, given ACC results can be interpreted based on what is known about the neurodevelopment of the primary auditory cortex, and CAEPs are reported to be affected by sleep stages (Lightfoot, 2016).
Although the ACC test sessions could be lengthy, it may be worth bearing in mind that any kind of audiological evaluation of young children can be a time-intensive ongoing process, particularly for infants (Boothroyd, 2009;Sabo, 1999).Example alternative behavioural methods for speech discrimination for very young children (from 6 months old) include VRISD and VRASPAC (visual reinforcement assessment of speech pattern contrasts) tests.Although some results may be obtained in a single test session (ca. 20-30 min;Boothroyd, 2009), more than one session may frequently be required to obtain accurate readings, and multiple test trials may be needed to obtain a conditioned head turn response for just a single data point (McCormick, 2004).And even then, the results from the behavioural measures may be unreliable (Cone, et al., 2022).Given this situation, and also that there are no standardised behavioural tests to obtain discriminative responses in very young children (Boothroyd, 2009), the ACC technique may prove a useful alternative to traditional methods despite the reported test times involved.
It is also worth considering the impact of the sound levels of the stimuli used to evoke ACCs, since these were frequently relatively loud (e.g.around 80 dB SPL for acoustic stimuli, Table 2), and how this may affect patient comfort levels.For instance, extended periods of loud sounds could be difficult for children with sensory processing issues or hyperacusis.
Data analysis of ACC measurements can be challenging due to the large amount of artefacts from myogenic noise, particularly in infants and young children finding it hard to sit still (Cone, et al., 2022).Multi-channel recordings may help improve signal to noise ratios, however, elaborate electrode montages required to acquire multi-channel data may be additionally time consuming requiring significant patient preparation (He, et al., 2015;Martin, et al., 2010), and there may be a playoff between the time required and potential resolution.In conclusion, it seems that efforts may be required to make the recording procedure as short, comfortable, and efficient as possible in order to facilitate the implementation of child ACC measurements in clinic.

Clinical applications of ACC measurements in children
Although the recording of ACC measurements in children is feasible, and experimental techniques exist, the question of possible clinical application arises.Results from the present review reveal that both the neural capacity for non-speech and speech sound discrimination in children can be studied by ACC measurements (Table 3).This is because a wide variety of stimuli were reported to elicit ACCs in children, encompassing investigations into temporal, frequency, intensity resolution, along with the ability to discriminate between a variety of short vowel and consonant speech sounds (Table 2, Table 3).Further, it was shown that such measures could be utilised in a variety of clinical populations, including children with SNHL, ANSD, and CAPD.Also, the review found that hearing interventions such as hearing aids, CI, and ABI did not present a barrier to ACC measurements, but rather ACCs may have the potential to help direct hearing intervention strategies (Table 3, Supplemental Material 6).Interestingly, given the age-dependence of ACC measurements, ACCs can also be used to investigate infants' phonetic vowel perception development (McCarthy, et al., 2019).

The performance of the ACC technique in children
A number of studies reported significant correlations between objective ACC measures and behavioural psychophysical threshold measures (Kumar, et al., 2020b;Nassar, et al., 2022) and speech discrimination (He, et al., 2013(He, et al., , 2015;;Xie, et al., 2022).Results, however, were sometimes mixed, and correlations were not always observed (He, et al., 2018;Petley, et al., 2024;Shalaby, et al., 2022;Xie, et al., 2022).In one example study, the type of speech material employed affected the results; sentence recognition in noise moderately correlated with ACC measures, whereas sentences in quiet did not (Xie et al. 2022).Another study showing mixed results when investigating possible correlations between behavioural and objective GDTs suggested that large variability between subjects may have clouded results (He, et al., 2018).When viewing the apparent variation between studies (e.g.He et al. (2015) observed a significant correlation between ACC measures and speech discrimination, whereas Shalaby et al. (2022) did not), it is important to consider the differences in listener groups (cf.Table 1).He et al. (2015) tested children with ANSD, a patient population with poor temporal processing acuity, attributable to poor neural synchrony (He et al. 2015).By contrast, Shalaby et al. (2022) tested children with typical SNHL; for these children, poor neural synchrony is not the primary pathophysiology underlying their hearing difficulties (Shalaby et al. 2022).Consequently, the different results reported are not necessarily contradictory.
In adult studies, some comparisons between ACC measures with behavioural measures of speech discrimination have also been reported.For example, a study investigating adult CI users found significant correlations between ACC measures (evoked by frequency changes) and speech perception results (van Heteren, et al., 2022).Another study found that ACCs significantly correlated with psychophysical measures for frequency and intensity discrimination, though, the ACC results were reportedly less sensitive than the behavioural measures (He, et al., 2012).Upcoming results from Biot et al. (2024) may help shed further light on the performance on the ACC, since this research group plans to compare ACC measurements to speech perception results for a total of 80 adult participants with varying degrees of SNHL (Biot, et al., 2024).
Overall, the performance of the ACC technique is promising.In future, it would be useful if more studies were conducted to examine the relationship between objective ACC measures and behavioural psychophysical tests in children, particularly comparing to perceptual measures of speech discrimination, since speech understanding and development is obviously important for children.Further, given the potential variability in ACC measures between subjects, larger sample numbers, with narrow age ranges (to avoid complications from maturational effects), may prove beneficial for such comparative performance investigations.

Study limitations
PRISMA guidelines recommend that a methodological quality assessment of included articles be conducted (Page, et al., 2021).A suitable standardised quality assessment tool was not identifiable from previously published literature.Hence the 'ACC Systematic Review Methodological Quality Assessment Checklist' (Supplemental Material 3) was derived, based on previously published similar tools (Downs and Black, 1998;Higgins and Green, 2008) for the purposes of this study.Although the methodological quality scores calculated here are not comparable externally to this study, they represent a quantitative intra-study metric, and help to provide insight into the overall quality of the included articles, whilst highlighting potential areas for improvement within this field of research.
In three of the studies (Kosemihal and Akdas, 2021;Shetty and Puttabasappa, 2020;Xie, et al., 2022) including teenagers (ages fifteen/sixteen years and above), the authors combined ACC data from the teenagers together with the adult results.Thus, although these three articles are compliant with the inclusion criteria based on age, it is difficult to separate out specific information related to ACC measurements on children from these particular studies (Kosemihal and Akdas, 2021;Shetty and Puttabasappa, 2020;Xie, et al., 2022).A number of other included studies, however, conducted ACC measurements specifically on older children and teenagers (Hamdy, et al., 2024;He, et al., 2013He, et al., , 2015He, et al., , 2014;;Kumar, et al., 2020aKumar, et al., , 2021;;Nassar, et al., 2022;Petley, et al., 2024).
Inter-subject variability in ACC responses was sometimes reported (He, et al., 2018;Martin, et al., 2010;McFayden, et al., 2020;Turkyilmaz, et al., 2013), and one hundred percent success rate in recording ACC responses was not always achieved for all the children tested (Martinez, et al., 2013;Strahm, et al., 2022).The reasons that ACC responses were not always obtained were unclear, and myogenic noise may sometimes have obscured results.It is also feasible that variations in maturation and refractoriness between individual infants may have contributed to differences in ACC waveforms.Perhaps freedom from the constraint of a limited testing time when assessing young children, or optimised stimuli parameters, may have unveiled additional ACC responses.Given the chances of recording a successful ACC response in a child are important to know from a clinical perspective, a greater clarity in the reporting of percentage ACC success rates in future publications could be beneficial in investigating these issues further.

Conclusions
ACC measurements are viable in both infants (from under three months old) and children.ACC measurements are age-dependent, and it is important to consider maturational effects.ACCs can be applied to study children with hearing loss, ANSD, CAPD, hearing aids, ABI and CI, and results may guide hearing rehabilitation strategies.ACC experimental procedures are often protracted; it would be beneficial to develop more efficient ACC recording protocols (e.g. by optimizing stimuli parameters).Further research is required to clarify the relationship between objective ACC responses and behavioural speech perception.Nevertheless, the ACC shows increasing promise for the assessment of the neural capacity for speech and sound discrimination in difficult-to-test hearing impaired children.

Fig. 1 .
Fig. 1.The PRISMA flow diagram of the study identification, the screening, the eligibility, and the inclusion process within the systematic search for acoustic change complex (ACC) measurements in children; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Table 2
Acoustic and Electrical Stimuli.The stimuli employed to evoke ACC responses in children.*Time reported for the EEG recording alone (i.e.time for preparation/set up time not included); **total time for ACC measurements including set up and preparation time.Abbreviations: CI, cochlear implant; CV, consonant vowel; IRN, iterated ripple noise; SNR, signal to noise ratio; VCV, vowelconsonant-vowel; VOT, voice onset time.

Table 2
(continued on next page) S.Meehan et al.

Table 2
(continued on next page) S.Meehan et al.