The inclusion criteria for the TD group are as follows: (1) individuals aged 6 years or older with no neurodevelopmental disorders, matched to the ASD group by age and sex; (2) individuals with no difficulties in verbal communication (eg, reading and writing), who provide written informed consent after receiving a full explanation of the study; for minors, written informed consent must be obtained from a legal representative after a sufficient explanation; and (3) males and females without any neurological or structural conditions that could impair neurodevelopment or cognitive function.

The inclusion criteria for the ASD group are as follows: (1) children or adults aged 6 years or older who have been diagnosed with ASD by a child/adolescent psychiatrist or a pediatrician based on the Diagnostic and Statistical Manual of Mental Disorders-Fifth Edition criteria; (2) individuals (or their legal representatives) who have received a full explanation of the study and have voluntarily agreed to participate by providing written informed consent; (3) participants whose symptoms cannot be better explained by other medical conditions or external factors; and (4) individuals deemed capable of successfully completing the study procedures.

The exclusion criteria for the TD group are as follows: (1) individuals with severe neuropsychiatric disorders; (2) a history of neurological or cerebrovascular diseases; (3) the presence of serious systemic medical conditions, such as hepatic, cardiac, or renal diseases; (4) a history of epilepsy or alcohol or drug abuse, or current use of psychotropic medications; (5) the presence of genetic disorders associated with developmental disabilities; and (6) individuals deemed unsuitable for the study at the discretion of the investigator.

The exclusion criteria for the ASD group were as follows: (1) individuals with cerebrovascular diseases such as cerebral hemorrhage or cerebral infarction; (2) individuals with developmental delay or severe cognitive impairment attributable to other primary medical conditions; (3) individuals with serious systemic medical conditions, such as hepatic, cardiac, or renal diseases; (4) a history of epilepsy, alcohol, or drug abuse; (5) individuals currently taking psychotropic medications for comorbid conditions who have had medication changes within the past month; and (6) individuals deemed unsuitable for the study at the discretion of the investigator.

Recruitment will be conducted through announcements on university websites and official posts. Participants will be recruited via the on-campus Integrated Support Center for Individuals with Developmental Disabilities and the Department of Korean Neuropsychiatry at the affiliated Korean Medicine Hospital. Additionally, recruitment will be conducted in collaboration with local communities, such as parenting forums and daytime activity support centers for individuals with developmental disabilities in the region. As the participants include minors, written informed consent will be obtained from their legal representatives. In the present study, although most participants are expected to be school-aged children and adolescents (elementary to high school level), adults with ASD may also be included; therefore, no strict age limitations are imposed. Given the male-to-female prevalence ratio of approximately 4:1 in patients with ASD, most participants are expected to be male. However, female participants will also be recruited whenever possible; therefore, sex is not a restricted criterion for inclusion.

A stratified blocked randomization with a block size of 4 will be used to assign participants in both the TD and ASD groups to the intervention or control group in a 1:1 ratio. Randomization will be conducted by an assigned researcher not involved in the assessment or intervention. Participants will be blinded to the type of intervention they receive. Throughout the study, the practitioner will not provide any clues to participants regarding group allocation. Blinding will be maintained throughout the study period. Following the second scan, participants (or their legal guardians) will be asked which intervention they believed they had received (verum or sham acupuncture) in order to assess the adequacy of participant blinding. Unblinding will be permissible only in exceptional circumstances in which knowing the participant’s allocated intervention is essential for their clinical management or safety. Specifically, unblinding may occur if a participant experiences a serious adverse event suspected of being associated with the study intervention. The decision to unblind will be made by the principal investigator after a thorough assessment of the situation. The date, time, and reason for unblinding, as well as the personnel involved, will be documented in detail in the participant’s case report form and reported to the institutional review board (IRB) within 24 hours.

For sample size estimation, a power calculation was performed using the G*Power software (version 3.1.9.4; Heinrich-Heine-Universität Düsseldorf) based on a 2-tailed t test. With an assumed effect size of 0.8, a significance level (α) of .05, and a power of 0.8, a minimum of 26 participants per group was required. Considering an anticipated dropout rate of 20%, a final sample size of 35 participants per group was deemed appropriate to ensure sufficient statistical power for both groups. The above power calculation was based on the primary comparison—namely, the difference in resting-state network metrics and relative powers in the frontal areas between the ASD and TD groups—for which the planned sample of 35 participants per group provides at least 80% power. As a sensitivity analysis, we also estimated the sample size required to detect the group × intervention interaction (analysis of covariance) within this stratified 2×2 design under a range of assumptions. Assuming a medium effect size (f=0.29), the required total sample size ranged from approximately 96 participants (pre-post correlation r=0.5) to 67 participants (r=0.7). This indicates that the interaction analysis is adequately powered if the pre-post correlation is high, or can otherwise be treated as exploratory given the current sample.

All participants will undergo assessments of vital signs (blood pressure, pulse rate, and temperature), demographic information, medical history (past, present, and family history), and current medication use. The K-CARS 2 will be administered for ASD screening at visit 0. Eligible participants meeting all inclusion and exclusion criteria will be scheduled for visit 1.

At visit 1, the investigator will monitor participants’ vital signs and any changes in their medication. The participants will undergo fNIRS and EEG scans twice, before and after acupuncture stimulation. Additionally, clinical indicators of ASD (SRS 2 and REVT) will be assessed to evaluate clinical status. Based on stratified randomization, participants will be assigned to 1 of 2 acupuncture stimulation groups. All acupuncture procedures will be performed by a licensed Korean Medicine doctor with over 3 years of clinical experience. Resting-state fNIRS will be recorded for 5 minutes, and resting-state EEG for 5 minutes, under eyes-closed and eyes-open conditions.

The investigator will administer the K-CARS 2 as an ASD screening tool to identify individuals with ASD and effectively distinguish them from those with other developmental disorders, while also measuring the severity of autistic symptoms [32]. Generally, a total score of 30 or higher (out of a maximum of 60) is used as the cut-off for ASD.

Additionally, clinical indicators of ASD symptoms will be evaluated using the SRS 2 and REVT. The SRS 2 is a 65-item parent-report questionnaire designed to quantitatively measure the continuum of typical and atypical social behaviors that covary with ASD symptom severity [33,34]. Each item is rated on a 4-point Likert scale that provides a quantitative measure of social impairment. The REVT will be used to assess language development, a key prognostic factor in ASD, focusing specifically on vocabulary through picture-based tasks [35]. The REVT assesses both receptive and expressive vocabulary abilities, providing standardized scores and percentiles to evaluate lexical development, and identifies potential language delays based on age-specific norms. To ensure objectivity and minimize bias, all clinical assessments will be conducted by an investigator blinded to the participants’ group assignments.

In the present study, NIRSIT (OBELAB) and SynAmps RT (Compumedics/NFEC-2024-11-300922) will be used for fNIRS and EEG data acquisition, respectively. Resting-state fNIRS data will be acquired for 5 minutes with eyes open, while resting-state EEG will be recorded under 2 conditions: 5 minutes with eyes closed and 5 minutes with eyes open. During the eyes-open recordings, participants will be asked to maintain a relaxed posture while gazing at a monitor. Both fNIRS and EEG data will be acquired before and after the intervention (Figure 2).

The fNIRS equipment comprises a portable neuroimaging modality designed to monitor cerebral oxygenation by delivering near-infrared laser light to the cortex. This system operates using 2 wavelengths, 780 nm and 850 nm, which are safe for human use and allow for the noninvasive measurement of hemodynamic changes. The NIRSIT system features 48 channels primarily focused on the frontal cortical regions, enabling the quantification of regional variations in oxygenated, deoxygenated, and total hemoglobin [36]. Accordingly, the analytical frameworks commonly used in fMRI research, such as graph theory and general linear model analyses, can be effectively implemented in fNIRS. To construct brain connectivity, the 48 frontal channels will be grouped into 8 regions of interest (ROIs) based on Brodmann areas, and an 8×8 connectivity matrix will be computed using Pearson correlation coefficients between ROIs (Figure 3).

EEG is an electroencephalography system used to record electrical signals in the brain by detecting neuronal voltage fluctuations. EEG measurements are based on electrical potential differences between electrodes on the scalp, which result from the propagation of current flow induced by synchronized postsynaptic potentials within neuronal membranes. The EEG system uses 32 channels to provide comprehensive coverage of the whole brain. In this study, a 12×12 coherence matrix will be constructed from 12 frontal channels (Fp1, Fp2, Fz, F3, F4, F7, F8, FCz, FC3, FC4, F11, and F12) selected from the 32 channels as ROIs (Figure 3).

Preprocessing and analysis of fNIRS data will be conducted using the NIRSIT Quest tool (v.1.3.1, OBELAB) and custom MATLAB code, whereas EEG data will be preprocessed and analyzed using Curry 9 (Compumedics Neuroscan) software together with the EEGLAB (v2026.0.0; Swartz Center for Computational Neuroscience, University of California San Diego) toolbox in MATLAB. The preprocessing procedures for fNIRS contain channel rejection, rejection padding, conversion to optical density, motion artifact correction, bandpass filtering, and conversion to hemoglobin concentrations [37]. The preprocessing of EEG includes data filtering, bad channel rejection, epoching, baseline correction, artifact removal using independent component analysis, interpolation, and rereferencing [38]. From the resulting connectivity matrices, network parameters, including local efficiency and clustering coefficient, will be computed using the Brain Connectivity Toolbox in MATLAB 2024b (MathWorks).

For each EEG channel, the power spectral density will be estimated using the Welch method (Hanning window, 50% overlap). Relative power will then be computed for the alpha (8‐13 Hz) and gamma (30‐45 Hz) bands as the ratio of the band power to the total power, and used for further analysis. By integrating these 2 modalities, comprehensive information regarding functional neural activity can be obtained from both hemodynamic responses and electrical oscillations of the brain.

In the study group, verum acupuncture stimulation will be conducted for 5 minutes using stainless steel acupuncture needles (0.20×30 mm, Dongbang Acupuncture) at specific acupoints previously shown to be associated with ASD symptom improvement, including Baihui (GV20), Yintang (EX-HN3), Sanyinjiao (SP6), Zusanli (ST36), Hegu (LI4), and Shenmen (HT7) [39]. To elicit a subtle deqi sensation, a mild manipulation of the needle will be applied, consisting of gentle lifting-thrusting (to a depth of 1‐2 cm) and twisting-rotation (90°‐180°), an approach considered appropriate and tolerable for pediatric individuals. After stimulation, a Deqi Sensation Questionnaire will be administered. For the control group, sham acupuncture will be performed using the same stainless-steel needles at nonacupoints unrelated to ASD symptoms, located within 1 cm of the corresponding verum acupoints on the head, face, and limbs. This will involve superficial needling with an insertion depth of only 1‐2 mm per point, followed by needle retention for 5 minutes. In this sham condition, the needles will remain static, and no attempt will be made to induce the deqi sensation.

Superficial needling is classified as a form of minimal acupuncture [40]. In the present study, the characteristics of the acupoints, depth and intensity of stimulation, and deqi sensation were considered the primary determinants of acupuncture efficacy. Therefore, minimal acupuncture, which lacks these features, was used as the sham control. However, this needling may still activate cutaneous mechanoreceptors, potentially eliciting minor nonspecific neural responses. Thus, the sham condition in this study should be considered an active control rather than an inert placebo. This form of sham-acupuncture has previously been established and applied in various clinical studies [41-44]. Furthermore, as this study involved minors with ASD, relatively small needles were selected to enhance participant compliance and ensure comfort throughout the research process.

This study aims to characterize disease-related differences in brain networks between the 2 groups, to elucidate the acute neural mechanisms of the acupuncture response, and to identify the specific neural substrates associated with the clinical characteristics of ASD. The primary outcome of this study will be defined as differences in resting-state frontal brain network parameters and relative powers of alpha and gamma bands in the frontal region according to disease status (ASD vs typical development). Secondary outcomes include differences in these neural markers between the 2 groups before and after acupuncture stimulation (verum vs sham acupuncture), reflecting the group × intervention interaction within this stratified 2×2 design. Exploratory outcomes will focus on identifying brain connectivity or network parameters associated with clinical indicators of ASD, such as the K-CARS 2, SRS 2, and REVT scores.

Demographic information (age, sex, height, and weight), medical history, and concomitant medications will be collected from each participant at study initiation. The investigators will record blood pressure, body temperature, respiratory rate, and pulse rate during visits 0 and 1. All adverse events will be comprehensively documented in case report forms, and the association between adverse events and interventions will be categorized as unrelated, possibly related, or related; severity will be classified as mild, moderate, or severe. In the event of serious adverse events, including death, life-threatening incidents, or events requiring hospitalization, the investigator will suspend part or all of the clinical trial and notify the coinvestigators and the IRB within 24 hours. Furthermore, as the trial may primarily involve minors, researchers will implement proactive measures to predict and mitigate the risks associated with potential noncompliance among minor participants.

Participants will be withdrawn from the study if neither fNIRS nor EEG can be performed, if a violation of the inclusion or exclusion criteria is discovered during the study, or if continuing the study becomes difficult owing to the development of adverse or serious adverse events.

Additional criteria for withdrawal include the discovery of a systemic disease not identified before study initiation, any significant violation of the study protocol, or withdrawal of consent by the participant. Furthermore, participants will be excluded if they take medications during the study period that could potentially affect the results, or in any other case in which the investigator determines that proceeding with the study is inappropriate.

Statistical analyses will include data from participants who have completed either fNIRS or EEG scans, along with clinical assessments of ASD. Categorical data will be presented as frequencies or percentages and analyzed using chi-square or Fisher exact tests. Numerical data will be reported as means and SDs, with comparisons conducted via independent t tests or Wilcoxon rank-sum tests, depending on normality.

To identify differences in frontal network patterns by disease status, 2-sample t tests will be conducted to compare network metrics (local efficiency and clustering coefficient) and relative band power between ASD and TD groups.

To examine the interaction effect of acupuncture within this stratified 2×2 design, a 2-way analysis of covariance will be conducted, with the postintervention values of each metric as the dependent variable and the corresponding preintervention values as a covariate. Disease group (ASD vs TD) and intervention type (verum vs sham) will be included as between-participants factors. Within this analysis, the group × intervention interaction will be the primary term of interest, indicating whether verum acupuncture modulates brain networks differently from sham depending on disease status. In addition, as an exploratory analysis, within-group paired t tests will be performed to examine pre-to-post changes in each disease group (ASD and TD), separately for the verum and sham conditions.

To investigate the neural substrates of ASD, partial correlation analyses will be conducted between connectivity and network metrics, and clinical scores (K-CARS-2, SRS-2, and REVT), with age and sex included as covariates.

For all brain connectivity analyses (t test and partial correlation), statistical significance will be assessed using both an uncorrected threshold of P<.001 and a false discovery rate–corrected threshold of P<.05 [45]. In contrast, network metrics and relative band power, both expressed as scalar values, will be evaluated at a conventional threshold of P<.05.

Statistical analysis will primarily be conducted according to the intention-to-treat principle, including all randomized participants who completed the baseline assessment. To address missing data, we will use multiple imputation techniques for clinical outcomes, provided the data are missing at random. For neuroimaging data, participants exhibiting excessive motion artifacts that remain uncorrectable after preprocessing will be excluded from the final analyses. Additionally, a per-protocol analysis will be performed as a sensitivity check to confirm the consistency of the primary findings.

The investigators will collect and record medical information on each participant’s case report form. As this study involves minors, informed consent will be obtained from both participants and their legal representatives. In addition, all study data, including patient confirmations, original signed informed consent forms, and detailed primary records, will be retained by the study institution for 3 years following the conclusion of the trial.

Important protocol modifications must be reported to the IRB. The initial approval period for the trial was 1 year, requiring regular reports and reapproval for subsequent periods. Prior to the study, all investigators involved in the intervention received comprehensive training in the clinical process. Two independent research assistants will crosscheck the data and case report forms to prevent errors. Currently, there are no plans for early termination of the study before December 2026. Once the target number of participants is reached, additional recruitment may proceed depending on the status of research funding.