ASD is identified as an early-onset neurodevelopmental disorder, the etiology of which poses a significant challenge on a global scale (1). Despite ongoing research efforts, the specific causes behind ASD remain largely undetermined, paralleled by a steady increase in diagnostic rates annually (2). Evidence supports early intervention as a pivotal strategy in ameliorating and alleviating the symptomatic expressions associated with ASD (3), underscoring the essence of precocious diagnosis or screening initiatives (4). Studies have shown that ASD symptoms typically manifest in early childhood, with 75-88% of children with ASD exhibiting signs within the first two years of life. The earlier the intervention, the more effective the treatment; research indicates that interventions before the age of three yield the most significant outcomes. However, in China, the median age for early ASD screening is 39 months, with a confirmed diagnosis typically occurring approximately one year after initial screening (5). Yet, empirical insights suggest that interventions administered prior to the completion of the third year of life yield superior outcomes (6), highlighting the critical need for the advancement of early screening timelines to effectively address ASD.

The decade subsequent to the DSM-5’s introduction has been characterized by a reiterated acknowledgment of perceptual abnormalities as a salient feature of ASD (7). Results from questionnaires administered to parents have shown that sensory perception (8), especially tactile sensitivity (9), serves as a typical marker of abnormalities. Moreover, some researchers believe that evidence from mouse models indicates that deficits in peripheral sensory neurons can contribute to ASD (10). Nevertheless, the assimilation of effective recognition and screening methodologies for these abnormalities remains insufficient within both clinical settings and the broader community context. An enhancement in awareness concerning these early markers is anticipated to facilitate the promotion of earlier screening and diagnosis endeavors, consequently enabling the provision of timely and efficacious interventions for individuals affected by ASD.

In recent years, some researchers have focused on quantifying tactile sensitivity issues in ASD using behavioral experiments or questionnaires. Tavassoli and colleagues (11) utilized a vibrotactile stimulation device to compare tactile sensitivity between individuals with ASD and TD participants. Participants were required to identify the stimulated finger after perceiving a weak vibration on either the index or middle finger. The results showed that higher tactile thresholds were significantly associated with more pronounced ASD traits. Anne and colleagues (12) developed a tactile frequency discrimination task using electrical stimulation. The ASD group demonstrated slower adaptability when adjusting to new stimulus frequency ranges, indicating that individuals with ASD differ from TD individuals in processing sensory inputs. This highlights the potential for incorporating tactile sensitivity into ASD auxiliary diagnostics. However, behavioral methods and questionnaires rely heavily on the cognitive abilities of participants. Consequently, these experiments are typically conducted on ASD individuals aged 18 and older, leaving a significant gap in the application of objective tactile sensitivity measures for early ASD screening, particularly for children under the age of 3.

Therefore, developing an evaluation paradigm that minimizes cognitive demands and does not require participants to perform tasks is critical for advancing research methodologies and improving early ASD detection. Electroencephalography (EEG) offers a promising approach in this regard. Recent studies have used EEG to investigate the neural specificity of individuals with ASD, primarily focusing on resting-state EEG (13, 14)and visual evoked potentials (VEP). Resting-state EEG assesses functional connectivity during rest, revealing atypical patterns in ASD, such as reduced low-frequency (delta, theta) connectivity and excessive high-frequency (beta, gamma) connectivity (13, 14), which correlate with ASD’s cognitive and behavioral traits. It is simple to conduct and suitable for all ages, including low-functioning individuals, but lacks task-related cognitive engagement, limiting its ecological validity. VEP (15), on the other hand, records brain responses to visual stimuli (e.g., faces, patterns) and highlights significant differences between ASD and TD individuals. ASD individuals often show atypical N170 waveforms in face processing tasks and abnormalities in P100 and P300 during visual pattern processing (15). While VEP effectively identifies perceptual and cognitive deficits, its reliance on visual stimuli and participant cooperation limits its applicability for individuals with visual impairments or attention deficits. Tactile paradigms bridge these limitations by requiring minimal cognitive effort while capturing task-related brain activity, offering a practical and objective alternative for studying cognitive processes in ASD. To date, research on tactile EEG in ASD remains extremely limited. The few existing studies predominantly focus on brain responses to simple tactile stimuli, often overlooking the assessment of tactile resolution. Notably, Piccardi et al.’s study employed a single-point tactile paradigm to investigate the neural markers of tactile sensory processing in 10-month-old infants at high risk for ASD or Attention Deficit Hyperactivity Disorder (ADHD). The neural response results revealed significantly reduced alpha wave desynchronization in high-risk ASD infants. These findings highlight the potential of tactile EEG paradigms for early identification of neural markers associated with ASD (16). However, the use of single-point, unvarying tactile stimuli does not allow for the investigation of tactile resolution, resulting in the largest tactile processing difference in ASD being overlooked. Thereby leaving a significant gap in the nuanced understanding of tactile processing and tactile resolution in ASD.

The current gaps in ASD tactile processing research are threefold: behavioral methods rely on cognitive abilities, limiting use in children under 3; EEG paradigms like resting-state EEG lack task engagement, and VEP depends on visual stimuli and cooperation; existing tactile EEG studies, such as Piccardi et al.’s, use simple stimuli, neglecting tactile resolution. Addressing these gaps requires developing objective, low-cognitive-demand tactile EEG paradigms capable of capturing multidimensional tactile processing characteristics, particularly for early ASD detection.

Previous research on the mechanisms of tactile event-related potential (ERP) has predominantly focused on typically developing individuals. Among the electrophysiological metrics, the ERP components associated with the brain’s processing of tactile stimuli include P100 (17, 18), N140 (19), P200 (20), N200 (21), and P300 (20) components are associated with the temporal processing of tactile stimuli in the brain. The P200 component, occurring during the mid-stage of tactile temporal processing, has been identified as a critical marker for distinguishing between multi-level tactile sensations. In contrast, the P300 component, observed during the late-stage of tactile temporal processing, exhibits a strong correlation with subjective judgments of tactile sensations (22). However, the electrophysiological performance of ASD in tactile temporal processing, especially in these components compared to TD, and the mechanisms of ASD in recognizing multi-level tactile stimuli, require further exploration.

The contribution of this study is its utilization of EEG to provide an objective assessment of tactile response resolution in individuals with ASD compared to TD counterparts, with the aim of elucidating distinct neurophysiological patterns and establishing tactile EEG as a robust physiological biomarker for ASD. In this study, we used a multi-channel tactile electrical stimulation and wireless EEG synchronization acquisition system suitable for ASD. We also developed a paradigm for multi-level pressure tactile stimuli. By sequentially analyzing ERP components, cortical activity, and the coefficient of variation (CV) in tactile information processing, we aimed to investigate cognitive deficits in ASD, particularly the dynamic information processing abnormalities during tactile recognition.

Autism’s social deficits are often associated with sensory abnormalities. Most previous studies have utilized visual and auditory stimuli as physiological biomarkers. However, increasing research also recognizes that abnormal tactile sensations are a significant reason why individuals with Autism Spectrum Disorder find it challenging to engage socially. Although some clinicians have suggested incorporating tactile assessments into clinical scales, implementing such measures remains difficult due to the lack of objectivity and reliability of questionnaires, particularly because most patients are children with developmental delays and mild cognitive impairments. Clinicians require a general benchmark of tactile response to clinically diagnose the extent of tactile abnormalities.

According to the results of our subjective survey, even typically developing children with normal intelligence have difficulty fully describing tactile sensations, let alone children with Autism Spectrum Disorder. Furthermore, there were no significant differences between the two groups in the questionnaire results. Therefore, it is challenging to objectively differentiate the tactile perceptions of children with Autism Spectrum Disorder and typically developing children based on questionnaires alone, making it particularly important to find an objective assessment method for early identification.

Previous studies have used methods such as MRI to objectively assess tactile neural processing in ASD. The study revealed that, compared to TD participants, individuals with ASD exhibited a typical modulation of connectivity between the sensorimotor regions and the prefrontal cortex during tactile stimulation (38). However, research on how the brain dynamically allocates resources to accomplish tactile discrimination tasks is still scarce, especially studies that objectively assess the abnormal tactile resolution in ASD children. In this study, we used a portable electro-tactile stimulation system that can be applied to ASD children and utilized EEG to objectively describe tactile processing in ASD and TD children. Through a series of EEG analyses and statistical comparisons, we aim to reveal the mechanisms of tactile processing deficits in ASD children.

Generally, normal adults process tactile sensations and form motor decisions through three stages (22, 25): the early stage (P100, N140) involves the right somatosensory association cortex distinguishing tactile types; the middle stage (P200) involves the right primary somatosensory cortex recognizing different degrees of tactile stimuli and forming tactile characteristics; the late stage (N200 and P300) involves information reaching the prefrontal cortex, where different objects are recognized based on individual cognition.

According to the performance shown in Figure 3 , TD children’s tactile processing can evoke typical tactile discrimination-related ERP components: P100–N140–P200–N200–P300. However, the early components in ASD children are not as typical as those in TD, and only P200, N200, and P300 components can be observed. On the other hand, statistical results indicate that the amplitudes of P200 induced at the PF2 lead, N200 induced at the F3 lead, and P300 induced at the C4 lead are significantly lower in ASD children compared to TD children, reflecting abnormalities in the mid-to-late stages of tactile processing in ASD, which implies that ASD has difficulties in differentiating tactile levels and forming tactile decisions. Combined with Figure 4A , we can identify the corresponding brain regions related to these mid-to-late stage abnormalities in the EEG components of ASD. It is evident that the brain regions recruited by ASD and TD differ across these three stages. The activation pattern in TD (late-stage C4 activation) is similar to the response to tactile stimulation reported in normal adults by Zhang et al. (21). In contrast, even in the late stages of tactile stimulation, the activation regions in ASD remain in the postcentral gyrus, an area typically activated in the early stages of tactile stimulation in normal adults (21). The above results suggest that information processing in the ASD group is delayed compared to the TD group. This phenomenon of sluggish tactile processing has also been reported in the behavioral experiments conducted by Katie et al., who found through behavioral studies that tactile brain processing in adult ASD individuals was delayed by 60 ms compared to adult TD individuals, while multisensory integration was delayed by 180 ms (39).

Interestingly, the abnormally activated regions observed during the tactile discrimination process in ASD (differences in brain regions FP2 and C4 for the five levels of tactile stimuli) are consistent with those identified in MRI studies (38), specifically in the prefrontal and sensorimotorareas, as illustrated in Figure 4B .

To further explore information processing during tactile recognition, particularly the efficiency of dynamic information processing, as illustrated in Figure 5 . we conducted a dynamic coefficient of variation analysis and performed pair-wise comparisons between the two groups (ASD vs. TD). We found that ASD children process tactile information throughout the entire time period; however, the coefficient of variation within a single time window is lower. This means that ASD children require more time but achieve lower efficiency to complete tactile cognitive tasks. Considering the entire brain spatial region, the coefficient of variation in the C4 of ASD children is lower than that of TD children over the entire time period, while it is higher in the FP2. This indicates that, in the context of low tactile processing efficiency in the C4, ASD patients exhibit compensatory tactile processing behavior in the FP2. This finding may reveal the defects in tactile recognition in ASD. Piccardi and colleagues (16)sought to identify physiological markers for ASD, exploring the use of tactile EEG as a potential biomarker. They emphasized the importance of early neural markers in predicting the later development of ASD, they innovatively proposed using tactile EEG as a physiological marker. However, their study did not conduct a comprehensive temporal analysis of tactile processing EEG, but instead relied on a single measure, the power spectral density at 300 ms. Additionally, their paradigm was not designed according to the most prominent tactile abnormalities in ASD (resolution issues), which might have resulted in fewer significant brain regions being identified.

In summary, results showed significantly reduced ERP amplitudes in ASD children at electrodes FP2, F3, and C4, indicating deficits in mid-to-late tactile processing stages. Brain topography revealed key group differences in regions such as the F8, FC5, FC6, T8, CP1, CP2, CP5, CP6, Pz, and Oz. CV analysis indicated prolonged but inefficient tactile processing in ASD, with compensatory activation in the FP2. These findings suggest that ASD children exhibit delayed tactile processing and inefficient discrimination of tactile stimuli. The identified EEG patterns and cortical activation differences provide a foundation for developing objective, early screening methods for ASD.

Traditional early screening methods for ASD (e.g., ABC, CARS, CHAT, ADOS, ESAT) primarily rely on parent-reported questionnaires (40), requiring approximately 30–60 minutes depending on questionnaire complexity and parental comprehension. In contrast, our paradigm, including preparation, is completed in approximately 15 minutes, improving the efficiency of early screening processes. Simultaneously, the integration of EEG enables the objective quantification of neural responses (41), providing a more precise and reliable approach to assessing brain activity. Wang and colleagues examined functional connectivity patterns in ASD using resting-state EEG. The study collected eyes-open resting-state EEG data from 72 children with ASD and 63 TD children and applied a data-driven clustering method to classify ASD into two subgroups: mild ASD (mASD) and severe ASD (sASD). The results revealed increased functional connectivity in the beta band for mASD and decreased connectivity in the alpha band for sASD compared to TD children, demonstrating that EEG can effectively distinguish between ASD and TD. However, the limitation of resting-state EEG lies in its focus on frequency characteristics, as it cannot capture temporal or spatiotemporal features, restricting its ability to fully explore dynamic neural activity in ASD. Marsicano et al. (42) utilized a visuo-spatial attentional task combined with EEG to investigate the dynamics of visual attention in individuals with ASD. The study included 19 children with ASD (mean age 11.21 years) and 20 TD children (mean age 11.25 years). Participants responded to visual targets following “zoom-in” or “zoom-out” cues. The results demonstrated prolonged neural encoding of visual cues in the ASD group, persisting even after target onset, whereas in the TD group, cue-related activity rapidly diminished after target appearance. This study also confirmed delays in sensory processing in ASD, similar to the tactile processing delays observed in our study. However, the visual task paradigm relied on participants making explicit judgments about visual stimuli and required significant cognitive engagement, limiting its applicability to younger children, particularly those under 3 years old, thereby reducing its potential utility for early screening applications. Tactile EEG paradigms demonstrate high efficiency, low demands on participants, and the ability to analyze multidimensional EEG information, highlighting their potential for early screening applications. Tactile electrical stimulation paradigms also present the challenge of potentially causing discomfort. In the future, more comfortable wearable flexible tactile systems will be designed to enhance user comfort.

This study demonstrated good test-retest reliability of EEG features during tactile tasks through ICC analysis. Specifically, the pressure-level CV values showed ICC=0.779 (95%CI:0.712-0.832) in the ASD group and ICC=0.729 (95%CI:0.653-0.792) in the TD group, which is largely consistent with the ICC range (0.68-0.81) reported by Zhang et al. (43) for tactile tasks. Using these stable features, we achieved a classification accuracy of 0.852 (95%CI:0.812-0.887) on the independent test set. Compared to recent similar studies: this result is slightly lower than the 0.89 accuracy obtained by Wang et al. (44) using multimodal data, but outperforms the 0.79 accuracy reported by Chen et al. (45) using resting-state EEG alone. Notably, the ERP component at pressure level 4 (P200 wave at Fz electrode) demonstrated both high reliability (ICC=0.751) and strong discriminative power (accuracy=0.852). This finding provides a biomarker with both stability and specificity for objective ASD diagnosis.

In future research, we will consider incorporating tactile discrimination tasks that reflect specific tactile behaviors in ASD, such as their aversion to dynamic touch and preference for static heavy pressing. However, it is important to acknowledge several limitations of this study. Firstly, despite extensive literature highlighting the gender disparity in autism prevalence, the participants in this study exhibit a gender imbalance. To address this, future research will recruit gender-matched subjects to evaluate the potential impact of gender on the findings. Additionally, future studies will explore the coupling and transmission of information between different brain regions, employing methods such as dynamic brain network analyses. IQ and language skills of participants were not assessed due to the age range of our sample, which included children as young as 1 year and 2 months, making formal IQ assessments infeasible. This lack of data may limit the interpretation of the relationship between cognitive abilities and tactile processing, To address this limitation, future studies will incorporate standardized IQ and language skill assessments, such as the Mullen Scales of Early Learning (MSEL) or the Peabody Picture Vocabulary Test (PPVT), depending on the age and cognitive abilities of the participants. Integrating these measures will allow for a more comprehensive analysis of how cognitive and linguistic factors may influence tactile processing in ASD. The statistical approach employed for analyzing questionnaire data may require further scrutiny. In future investigations, we intend to incorporate qualitative analytic techniques—such as thematic analysis—and to recruit larger sample sizes, thereby enhancing the robustness and generalizability of our findings. Furthermore, correlating these cognitive and language assessments with EEG-based classification results could provide additional insights into the neural mechanisms underlying tactile processing and its relationship with broader developmental profiles in ASD. While our pressure levels were initially determined from healthy adult thresholds for experimental consistency, Future studies should incorporate such pediatric-specific calibration, particularly when comparing absolute sensitivity thresholds between groups. Designing experimental paradigms targeting these specific tactile behaviors will help more accurately identify significantly abnormal brain regions and provide a more comprehensive understanding of tactile processing in ASD. Once the mechanisms underlying tactile processing abnormalities in ASD are fully understood, future research will focus on recruiting a large number of participants to build an EEG feature database for tactile processing. Using artificial intelligence algorithms, we will classify the EEG features of ASD and TD individuals, aiming to develop a binary classification method for early screening, which could be applied to intelligent diagnostic tools for clinical early screening.

This study employed EEG-based analysis to investigate the neural mechanisms underlying tactile processing differences between children with ASD and TD children. The results revealed distinct neurophysiological patterns in ASD, characterized by significantly reduced amplitudes of mid-to-late ERP components (P200, N200, and P300) at key electrodes (FP2, F3, and C4), indicating impaired tactile discrimination and decision-making processes. Brain topography further highlighted group differences, with ASD children exhibiting atypical activation in regions such as F8, FC5, FC6, T8, CP1, CP2, CP5, CP6, Pz, and Oz, suggesting compensatory neural recruitment despite overall inefficiency in tactile processing. CV analysis demonstrated prolonged but less efficient tactile information processing in ASD, with compensatory activity observed in the FP2. These findings underscore the potential of EEG-derived biomarkers, such as ERP components and CV values, for objective early screening of ASD. The study also validated the reliability of these EEG features, with machine learning models achieving high classification accuracy, supporting their utility in clinical applications. In summary, this research provides novel insights into the neural basis of tactile processing abnormalities in ASD and lays the groundwork for developing non-invasive, objective diagnostic tools. Future studies should expand sample sizes, incorporate longitudinal designs, and explore multimodal approaches to further refine these biomarkers and enhance their clinical applicability.