Cluster analysis based on sensory characteristics of children with autism spectrum disorder and its impact on emotional and behavioral manifestations and parenting stress Authors: Zhanbin Xu, Feiying Wang, Hongchao Qin, Hao Jiang, Libing Ma, Qiuyang Zhang, Jia Wang, Lijie Wu Published: 2026 ABSTRACT AbstractAutism spectrum disorder (ASD) exhibits substantial sensory processing heterogeneity, linked to diverse emotional-behavioral manifestations and caregiver challenges. Sensory-based subgroups may reveal distinct clinical profiles and inform targeted interventions, yet their associations with parenting stress remain underexplored. While prior work has examined sensory features, emotional-behavioral manifestations, and parenting stress in autistic children, few studies have characterized sensory subgroups and their correlates. This study employed K-Means cluster analysis combined with the elbow method to categorize autistic children into distinct subgroups based on the three The Short Sensory Profile (SSP) dimensions with the highest variability, aiming to explore the sensory characteristics of different subgroups and their impacts on emotional and behavioral manifestations and parenting stress. The research included 225 autistic children and 228 comparison group participants, using SSP, The Strengths and Difficulties Questionnaire (SDQ) and The Parenting Stress Index-Short Form (PSI-SF) to assess sensory features, emotional and behavioral manifestations and parenting stress in both groups. The results showed that the three dimensions with the highest proportions of sensory differences in children with ASD were auditory filtering (the capacity to filter out irrelevant background sounds) had the highest rate at 68.44%, followed by low energy/weak (66.67%), and movement sensitivity (51.56%), which were significantly higher than those in the control group (26.32%, 24.56%, 27.19%) (all P<0.01). Using the K-Means clustering algorithm combined with the elbow method, children with ASD were classified into three subgroups: Group 1 as the “sensory adaptive type”, Group 2 as the “motor-sensitive/weakness type” and Group 3 as the “auditory filtering type”. Structural equation modeling results showed no significant association between sensory features and emotional and behavioral manifestations or total parenting stress scores in Group 1 (all P > 0.05). In Group 2, auditory filtering had a significant direct negative effect on both the SDQ total difficulties score and PSI-SF total parenting stress score (β = -0.25, -0.38, P < 0.05), while movement sensitivity had a significant direct negative effect on the SDQ total difficulties score (β = -0.26, P < 0.05). In Group 3, auditory filtering exerted a significant direct negative effect on the PSI-SF total parenting stress score (β = -0.25, P < 0.01) and also indirectly influenced it through the SDQ total difficulties score (β = -0.22, 0.31, P < 0.05). Additionally, sex had a significant direct effect on the total parenting stress scores of the PSI-SF in group 3 of children with ASD (β = 0.17, P < 0.05). Sensory differences are prevalent in ASD and correlate with emotional/behavioral manifestations and parenting stress. However, different types of sensory profiles influence emotional/behavioral characteristics and parenting stress through distinct pathways. Compared to the “sensory adaptive type” and “motor-sensitive/weakness type” subgroups, the “auditory filtering type” subgroup not only exerted a significant direct negative effect on parenting stress but also demonstrated a significant indirect negative effect through the mediating role of emotional/behavioral manifestations. Early identification of distinct ASD subgroups using sensory features may facilitate the timely development of targeted intervention plans and support policies, which improve clinical symptoms in children with ASD and alleviate parenting stress.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12888-026-08259-8.Keywords: Autism spectrum disorder, Sensory features, Subgroups, Cluster analysis, Emotional and behavioral manifestations, Parenting stress Introduction Autism spectrum disorder (ASD) is a neurodevelopmental condition that manifests in early childhood, typically emerging before the age of 3. It is characterized by persistent impairments in initiating or maintaining social interactions, accompanied by restricted interests, repetitive stereotyped behaviors, and sensory differences [1–3]. The etiology and mechanisms of ASD remain unclear, and its clinical features exhibit significant heterogeneity, making precise diagnosis and targeted intervention challenging. ASD is a prevalent neurodevelopmental condition in childhood, characterized by persistent neurodivergent traits that may require long-term support [4, 5]. In recent years, owing to its high prevalence and related impact on individuals and families, ASD has garnered extensive attention from both national authorities and society at large [6, 7]. Sensory differences are a core feature of ASD newly included in under-responsive [8]. Although their mechanisms remain unclear, the degree of sensory differences can predict clinical symptoms and prognosis in children with ASD, such as social difficulties, emotional and behavioral manifestations [9, 10]. The decision by the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) to consolidate several related disorders into a single diagnosis of ASD has sparked significant controversy and debate, as various ASD subtypes differ in characteristics and symptom intensity [11, 12]. Currently, sensory differences as a core feature of ASD receive insufficient attention in both clinical practice and research. Research on the sensory characteristics of children with ASD has predominantly focused on atypical manifestations within single sensory modalities or employed overall scoring methods to analyze sensory profiles, overlooking the heterogeneity of these characteristics [13]. Different children with ASD may exhibit distinct combinations of sensory patterns [14]. However, existing studies remain relatively scarce in their attention to and exploration of clustering differences in sensory characteristics among children with ASD. Notably, ASD subgroups with different sensory clusters demonstrate marked variations in clinical symptoms such as social interaction, emotional regulation, and behavioral manifestations [15–17]. Studies indicate that children with ASD often exhibit significant emotional and behavioral manifestations, such as heightened emotional reactivity and poor emotional regulation [18], sleep disturbances [19], hyperactivity/attention deficits [20], and difficulties in peer interactions [21], which can lead to severe social maladjustment and functional impairments [22]. Parents of children with ASD frequently express concerns about their children’s behavioral manifestations, daily living skills, and ability to manage emotions and behaviors during the parenting process [23]. Early interventions targeting emotional and behavioral manifestations can effectively improve outcomes [24], with approaches such as cognitive behavioral therapy [25], family-based behavioral interventions [26, 27], exercise interventions [28], and dietary restrictions [29] demonstrating efficacy in modulating behavioral manifestations. The emotional and behavioral challenges of children with ASD often hinder families from seeking appropriate therapeutic interventions, making it necessary to consider parental perspectives when developing intervention strategies [30]. Research indicates that caregivers of autistic children consistently experience higher levels of parenting stress during early childhood development, compared to caregivers of children with non-autism-related developmental concerns and neurotypical children [31–33]. While such stress is often tied to children’s emotional and behavioral presentations, it varies widely across individuals and does not consistently correlate with autism symptom severity [34]. Studies indicate that many parents of children with ASD experience poor coping abilities, which are associated with reduced quality of life relative to parents of typically developing children [35]. Longitudinal studies further suggest a potential bidirectional predictive relationship between emotional and behavioral manifestations in children with ASD and parental psychological stress [36]. Parents of children with ASD experience greater parenting stress and more mental health issues compared to parents of typically developing children [37]. However, the extent to which these intertwined family psychosocial experiences are related has received little research attention [38]. Furthermore, parenting stress is shaped by multiple factors, including child behavioral and sensory subtypes, coping strategies, and social and policy support, with stress communication and dyadic coping identified as key moderators of its impact on family adjustment [39]. In particular, variations in behavioral and sensory profile subtypes may further exacerbate caregiver stress [40, 41], while systemic barriers to services and lack of social and policy support can increase anxiety and depressive symptoms in these parents [42, 43]. As outlined earlier, parenting stress among caregivers of autistic children is shaped by multiple factors, including children’s sensory and behavioral characteristics. Early sensory differences in autistic children are known to correlate with emotional and behavioral manifestations and higher parenting stress [44, 45], yet there remains a critical gap in understanding how distinct sensory profile subgroups relate to these outcomes. To address this gap, the present study evaluates sensory characteristics, emotional and behavioral manifestations, and parenting stress in autistic children and neurotypical peers. Using k-means cluster analysis focused on the most clinically relevant dimensions of the SSP, we identify meaningful sensory subgroups and explore differences in emotional and behavioral manifestations and parenting stress across groups. We further examine both the direct impact of sensory characteristics on parenting stress and the indirect impact mediated by emotional and behavioral manifestations within each subgroup. The findings aim to inform targeted early intervention strategies and support policies to address emotional and behavioral manifestations and alleviate parenting stress. Research subject A total of 225 children with ASD who received rehabilitation therapy at designated autism rehabilitation institutions in Nantong from June 2023 to February 2024 were enrolled as the observation group (ASD group), including 166 boys (73.78%) and 59 girls (26.22%), with a male-to-female ratio of 2.81:1 and a mean age of 6.46 ± 2.06 years. All autistic children were diagnosed by psychiatrists or pediatricians according to the diagnostic criteria for autism spectrum disorder in the DSM-5 [1], while excluding other severe organic or psychiatric disorders. Additionally, 228 children who underwent physical examinations at Nantong Maternal and Child Health Hospital during the same period were selected as the comparison group (TD group), consisting of 171 boys (75.0%) and 57 girls (25.0%), with a male-to-female ratio of 3.0:1 and a mean age of 6.27 ± 2.11 years. There were no statistically significant differences in age or sex composition between the two groups (t-value and χ2 value were 0.589 and 0.089, respectively, with P-values of 0.556 and 0.766), indicating comparability. Investigation method This study employed a cross-sectional survey method. From June 2023 to February 2024, 225 children with ASD receiving training at designated autism rehabilitation institutions in Nantong City were enrolled as the observation group (ASD group), while 228 children undergoing physical examinations at Nantong Maternal and Child Health Hospital during the same period were selected as the comparison group (TD group). The SSP was used to assess sensory characteristics in both groups. The SDQ and PSI-SF were administered to evaluate emotional/behavioral manifestations in children and parenting stress levels respectively. All researchers involved in this study received standardized training to ensure familiarity with the study procedures and proficiency in using the assessment tools. The researchers provided on-site guidance to parents for questionnaire completion, conducted real-time quality control, and designated personnel were responsible for data collation and analysis after collection. Measuring tools and methods The short sensory profile (SSP) SSP is one of the most commonly used questionnaires for assessing children’s sensory characteristics [46]. It is suitable for screening sensory processing differences in children aged 3 to 10 years. The results include seven dimensions—tactile sensitivity (response patterns to tactile stimuli, including clothing texture, physical touch, wet and dry contact), taste/smell sensitivity (responses to gustatory and olfactory stimuli), movement sensitivity (responses to vestibular and proprioceptive activities), under-responsive/seeks sensation (behavioral tendency of under-responsiveness to sensory input or active seeking of intense sensory stimulation), auditory filtering (the capacity to filter out irrelevant background sounds), low energy/weak (overall arousal level and physical energy condition), and visual/auditory sensitivity(response patterns to visual and auditory stimuli such as bright light, flicker and noise)—along with a total sensory score, comprising 38 items in total. Based on the frequency of behaviors described in the items, a 1–5 rating scale is used. Scores for each dimension and the total score are categorized into three levels: “typical performance”, “probable difference” and “definite difference”. For the SSP total score, the classification ranges are 155–190 for typical performance, 142–154 for probable difference, and 38–141 for definite difference. Each sub-dimension adopts its own independent normative cutoff ranges as defined in the official scale manual. Lower scores in each dimension, the total score, and their standardized mean values indicate more pronounced sensory differences. The SSP has been widely used internationally and in China [47, 48], and has been extensively applied to sensory assessment among autistic children in China [48, 49], with good psychometric applicability and acceptable reliability and validity. The Cronbach’s α coefficient was 0.90 in this study. The strengths and difficulties questionnaire (SDQ) SDQ is a commonly used scale for assessing children’s emotional and behavioral manifestations [50, 51]. It is completed by primary caregivers based on the child’s performance over the past six months. The scale consists of 25 items, each rated on a 0–2 point scale: 0 (not true), 1 (somewhat true), and 2 (certainly true). It evaluates five factors: emotional symptoms, conduct problems, hyperactivity/inattention, peer relationship problems, and prosocial behavior, as well as a total difficulties score. Higher scores on emotional symptoms, conduct problems, hyperactivity/inattention, peer relationship problems, and total difficulties, as well as lower prosocial behavior scores, correspond to elevated emotional and behavioral manifestations. The SDQ has been widely used internationally and in China [52, 53], and has been extensively applied to behavioral assessment among autistic children in China [29, 54], with good psychometric applicability and acceptable reliability and validity. The Cronbach’s α coefficient was 0.784 in this study. The parenting stress index-short form (PSI-SF) PSI-SF is a commonly used tool for assessing parenting stress [55, 56]. The scale consists of 36 items rated on a 5-point Likert scale (1 indicating strong disagreement, 5 indicating strong agreement), with higher total scores indicating more severe parenting stress. The PSI-SF total score ranges from 36 to 180 points, with 90 points set as the cutoff score. A total score above 90 indicates that parents are experiencing high levels of stress during child-rearing, while a score of 90 or below suggests low levels of parenting stress. Based on the Parenting Stress Index score results, the levels were categorized into four tiers: normal level (≤ 85), borderline high level (86–90), high level (91–98), and very high level (≥ 99). The PSI-SF has been widely used internationally and in China [55, 57], and has been extensively applied among autistic children in China [58, 59], with good psychometric applicability and acceptable reliability and validity. The Cronbach’s α coefficient was 0.935 in this study. Analytical methods Based on the theoretical premise that children with ASD exhibit heterogeneous sensory profiles with potential subtype differences in emotional-behavioral manifestations and parenting stress [41, 60], this unsupervised clustering approach delineated sensory-based subgroups to provide a data-driven foundation for subsequent inter-subgroup comparative analyses. For clustering, we employed a consensus k-means approach with bootstrap resampling to enhance the stability and reproducibility of cluster assignments. The elbow method was adopted to determine the optimal cluster number as a standard heuristic for data partitioning [15, 61]. All analyses were performed in R using the ConsensusClusterPlus package to reduce the susceptibility of conventional K-Means to random initial configurations and sampling fluctuations. Clustering was performed on three sensory dimensions with the highest variability, namely auditory filtering, low energy/weak, and movement sensitivity, to identify distinct sensory processing patterns. These three dimensions were chosen for their highest sample variability to effectively distinguish sensory subgroups. This selection ensures clear clinical implications, supporting targeted sensory intervention and behavioral guidance. We further conducted clustering analyses with 4, 5, 6 and all seven dimensions for sensitivity verification. Compared with models including more redundant indicators, the three-dimension approach presents more stable clustering performance, preserves core sensory characteristics and simplifies phenotypic interpretation. Data description and analysis were conducted using SPSS 26.0 software. Normally distributed data were expressed as mean (± s). Categorical variables were presented as counts (n) and corresponding percentages (%). Intergroup comparisons were performed using t-tests, ANOVA, and χ2 tests. Exploring the association between sensory features and emotional and behavioral manifestations in children with ASD and parenting stress through structural equation modeling. Structural equation modeling (SEM) was conducted using the lavaan package in R software [62]. The maximum likelihood with robust standard errors (MLR) estimator was employed for model fitting, with a significance level set at α = 0.05. Result Comparison of general characteristics and SSP, SDQ, PSI-SF results between ASD and TD group children The study results showed that children in the ASD group scored significantly lower than the TD group in all SSP dimensions, including tactile sensitivity, taste/smell sensitivity, movement sensitivity, low energy/weak, auditory filtering, under-responsive/seeks sensation, and visual/auditory sensitivity, as well as in the total score (P < 0.001). As defined in the Methods section, lower SSP scores reflect more pronounced sensory processing differences, indicating greater sensory challenges in the ASD group relative to the TD group. The SDQ results indicated that autistic children scored higher than the TD group in emotional symptoms, conduct problems, hyperactivity/inattention, peer relationship problems, and total difficulties, while scoring lower in prosocial behavior (P < 0.001). Compared with TD children, autistic children showed more pronounced emotional and behavioral manifestations. The PSI-SF results revealed that parents of autistic children scored significantly higher than those of TD children in parenting stress (PD), parent-child dysfunctional interaction (PCDI), difficult child (DC), and total parenting stress (TD) (P < 0.001). Compared to TD children, parents of children with ASD experienced higher levels of parenting stress. Table 1. Table 1Comparison of general characteristics and SSP, SDQ, PSI-SF results between ASD and TD group childrenDimensionASD groupTD groupt/χ2 P n meanSDRange n meanSDRange AGE 2256.462.063–102286.272.113–100.5890.556 SEX Male1661710.0890.766Female5957 Short sensory profile 225228Tactile sensitivity31.114.1614–3533.302.4022–35-6.890<0.001Taste/Smell sensitivity14.974.274–2017.772.669–20-8.393<0.001Movement sensitivity12.072.673–1513.392.006–15-5.988<0.001Under-responsive/seeks sensation26.884.8512–3531.963.2320–35-13.146<0.001Auditory filtering19.675.356–3025.223.9111–30-12.619<0.001Low energy/weak22.245.916–3027.313.5013–30-11.119<0.001Visual/auditory sensitivity19.503.787–2521.713.0113–25-6.910<0.001 Total 146.4322.7477–190170.6716.37124–190-13.035<0.001 SDQ 225228Emotional symptoms2.701.880–81.601.480–86.936<0.001Conduct problems2.431.490–81.461.050–58.011<0.001Hyperactivity/inattention6.252.300–103.362.170–1013.757<0.001Peer relationship problems4.711.910–91.431.340–621.116<0.001Prosocial behavior4.002.600–107.921.743–10-18.857<0.001 Total difficulties 20.095.039–3515.763.598–3210.537<0.001 PSI-SF 225228Parental distress34.159.4112–5925.357.8012–5210.826<0.001Parent–child dysfunctional interaction32.048.6512–5620.896.2612–4215.737<0.001Difficult child33.488.9812–6023.857.1312–4412.657<0.001 Total distress 99.6723.7836–17570.0818.6338–12814.751<0.001 Comparison of general characteristics and SSP, SDQ, PSI-SF results between ASD and TD group children Under-responsive/ seeks sensation Comparison of sensory differences between ASD and TD groups in children The results showed that the ASD group had significantly higher variability across all SSP dimensions compared to the TD group (P < 0.001). The three dimensions with the highest rates of sensory differences in autistic children were auditory filtering (68.44%), low energy/weak (66.67%), and movement sensitivity (51.56%), with significantly higher rates than the TD group (26.32%, 24.56%, 27.19%) (P < 0.001). These were followed by under-responsive/seeks sensation (44.0%), taste/smell sensitivity (41.78%), visual/auditory sensitivity (36.0%), and tactile sensitivity (27.11%), all showing significantly higher rates than the TD group (7.46%, 11.40%, 15.79%, 7.89%) (P < 0.001). See Fig. 1. Comparison of the percentages of children with sensory processing differences (i.e., scores in the probable or definite difference range) in each SSP domain, by group. Percentages were calculated within each group as: (Number of children with sensory processing differences in the domain / Total number of children in the group) × 100%. Under/Seeks*, under-responsive/seeks sensation; Aud/Filt*, auditory filtering; LEW*, low energy/weak; Vis/Aud*, visual/auditory sensitivity. *** p < 0.001 Cluster analysis of sensory characteristics in children with ASD As shown in Fig. 1, the three dimensions with the highest variability were auditory filtering, low energy/weak, and movement sensitivity (all with difference rates > 50%, as observed in this study). A k-means cluster analysis was applied to analyze the three highest-variability dimensions among 225 autistic children. Bootstrap sampling was performed with K ranging from 3 to 10 for 500 iterations of k-means clustering, and the consensus clustering for each K was used as the final clustering for subsequent analysis. According to the elbow method, the optimal clustering effect was achieved when K = 3. The elbow plot and clustering effect diagrams are presented in Figs. 2 and 3. Visualization of the “elbow rule” used to determine the optimal number of clusters in k-means clustering. The plot shows a distinct “elbow” feature when k = 3. According to the elbow rule, the optimal number of clusters for this dataset is 3 Visualization of k-means clustering results using PCA dimensionality reduction (K = 3). The first principal component (Dim1) explains 63.8% of the variance, and the second (Dim2) explains 21.0%, together accounting for 84.8% of the total variance. Each point represents an individual child with ASD Comparison of sensory characteristics among different subgroups of children with ASD The 225 children with ASD were divided into three groups through k-means cluster analysis based on three dimensions: movement sensitivity, auditory filtering, and low energy/weak. Standardized mean values in the cluster analysis are directly derived from raw SSP scores. Specifically, higher standardized mean values correspond to higher raw scores, which align with the typical sensory processing range in the SSP manual. Lower standardized mean values correspond to lower raw scores, which fall into the probable or definite difference ranges. Group 1 (33.33%) showed higher standardized mean values across all three dimensions, with auditory filtering > 1.0, and both low energy/weak and movement sensitivity > 0.5. Children in group 1 with ASD demonstrated typical sensory processing in the three dimensions of movement sensitivity, auditory filtering, and low energy/weak; this group was designated as the “sensory adaptive type”. In Group 2 (29.78%), autistic children showed mean values <-1 for both movement sensitivity and low energy/weak features, and <-0.5 for auditory filtering. This group was named the “motor-sensitive/weakness type”. In Group 3 (36.89%), autistic children had mean values between 0 and 0.5 for both movement sensitivity and low energy/weak dimensions, with a mean value of -0.5 for the auditory filtering dimension. Group 3 was designated as the “auditory filtering type”. Moreover, the study results showed statistically significant differences (P < 0.001) in both intergroup and intragroup comparisons among the three groups of autistic children across three dimensions: sensory sensitivity related to movement sensitivity, auditory filtering and low energy/weak. Figures 4 and 5 present the heatmap and bar chart of standardized feature means for different subgroups, respectively. The standardized feature means heatmap of three clusters, indicating the differences in feature means among three groups of autistic children across auditory filtering, movement sensitivity, and low energy/weak in three sensory dimensions. Red represents high feature means in different clusters, while blue represents low feature means Bar chart of standardized feature means for three clusters, indicating significant differences in mean feature values among three groups of autistic children across auditory filtering, low energy/weak, and movement sensitivity sensory dimensions, both between and within groups, *** P < 0.001 The correlation between sensory differences in children with ASD and their emotional and behavioral manifestations as well as parenting stress The results showed that the scores of each dimension and the total score of SSP in children with ASD were negatively correlated with emotional symptoms, conduct problems, hyperactivity/inattention, peer relationship problems, and total difficulties score of SDQ, while positively correlated with prosocial behavior (P < 0.05). The scores of each dimension and the total score of SSP were negatively correlated with parenting stress (PD), parent-child dysfunctional interaction (PCDI), difficult child dimensions (DC), and total parenting stress score (TD) of PSI-SF (P < 0.01). Emotional symptoms, conduct problems, hyperactivity/inattention, peer relationship problems, and total difficulties score of SDQ were positively correlated with each dimension and total score of PSI-SF, while prosocial behavior was negatively correlated with each dimension and total score of PSI-SF (P < 0.01). See Tables 2 and 3. Table 2Correlation analysis between SSP dimensions and SDQ, in children with ASD and PSI-SFSSP DomainsEmotional symptomsConduct problemsHyperactivity/inattentionPeer problemsProsocial behaviorTotal difficultiesPDPCDIDCTDTactile sensitivity-0.33***-0.38***-0.20***-0.25***0.23**-0.30***-0.31***-0.38***-0.42***-0.42***Taste/smell sensitivity-0.21**-0.11*-0.31***-0.30***0.27***-0.23**-0.24***-0.28***-0.36***-0.33***Movement sensitivity-0.39***-0.24***-0.24***-0.25***0.15*-0.34***-0.26***-0.33***-0.36***-0.36***Under-responsive/seeks sensation-0.44***-0.27***-0.47***-0.42***0.38***-0.42***-0.28***-0.41***-0.46***-0.43***Auditory filtering-0.32***-0.20**-0.62***-0.52***0.41***-0.45***-0.27***-0.43***-0.46***-0.44***Low energy/weak-0.36***-0.19**-0.31**-0.26***0.28***-0.29***-0.24***-0.27***-0.24***-0.29***Visual/auditory sensitivity-0.42***-0.33***-0.30***-0.26***0.19**-0.40***-0.23**-0.33***-0.37***-0.35***Total-0.48***-0.33***-0.50***-0.46***0.39***-0.48***-0.35***-0.47***-0.52***-0.51****P<0.05, **P<0.01, ***P<0.001. Significance levels are unadjusted for multiple comparisons Correlation analysis between SSP dimensions and SDQ, in children with ASD and PSI-SF Under-responsive/ seeks sensation *P<0.05, **P<0.01, ***P<0.001. Significance levels are unadjusted for multiple comparisons Table 3Correlation analysis between SDQ dimensions in children with ASD and PSI-SFPSI-SF DomainsEmotional symptomsConduct problemsHyperactivity/inattentionPeerproblemsProsocial behaviorTotal difficultiesPD0.306***0.205**0.222**0.227**-0.193**0.263***PCDI0.328***0.267***0.379***0.397***-0.3560.342***DC0.436***0.427***0.380***0.449***-0.300***0.479***TD0.405***0.340***0.369***0.404***-0.319***0.409****P<0.05, **P<0.01, ***P<0.001. Significance levels are unadjusted for multiple comparisons Correlation analysis between SDQ dimensions in children with ASD and PSI-SF *P<0.05, **P<0.01, ***P<0.001. Significance levels are unadjusted for multiple comparisons Comparison of emotional and behavioral manifestations and parenting stress in different subgroups of children with ASD The study results showed significant differences in emotional and behavioral manifestations and parenting stress among different ASD subgroups (P < 0.01). Group 2 autistic children scored highest on SDQ emotional symptoms, conduct problems, hyperactivity/inattention, peer relationship problems, and total difficulties scores, followed by Group 3 and Group 1 (P < 0.01). Group 2 had the lowest prosocial behavior scores, followed by Group 3 and Group 1 (P < 0.01). Regarding PSI-SF dimensions, Group 2 autistic children scored highest on parenting stress, parent-child dysfunctional interaction, difficult child characteristics, and total parenting stress scores, followed sequentially by Group 3 and Group 1 (P < 0.01). Specific details are presented in Table 4. Table 4Comparison of SDQ and PSI-SF dimension results among different clustering groups of autistic childrenDimensionCluster 1(n = 75)Cluster 2(n = 67)Cluster 3(n = 83)F P ηp2Significant post-hoc comparisons SDQ Emotional symptoms1.81 ± 1.483.64 ± 1.832.75 ± 1.8819.572<0.0010.15C2 > C1***; C2 > C3**;C3 > C1**Conduct problems1.93 ± 1.492.81 ± 1.642.57 ± 1.237.0060.0010.06C2 > C1**;C3 > C1*Hyperactivity/inattention4.64 ± 2.027.10 ± 2.157.01 ± 1.8636.337<0.0010.25C2 > C1***;C3 > C1***Peer problems3.63 ± 1.755.39 ± 1.855.13 ± 1.6921.656<0.0010.16C2 > C1***;C3 > C1***Prosocial behavior5.03 ± 2.643.07 ± 2.573.83 ± 2.2711.226<0.0010.09C2 > C1***;C3 > C1**Total difficulties17.04 ± 4.0422.01 ± 5.2721.29 ± 4.3725.701<0.0010.19C2 > C1***;C3 > C1*** PSI-SF PD 31.56 ± 9.0637.78 ± 9.5033.55 ± 8.838.524<0.0010.07C2 > C1***; C2 > C3* PCDI 27.45 ± 7.0435.58 ± 8.8633.31 ± 8.0719.967<0.0010.15C2 > C1***;C3 > C1*** DC 28.85 ± 7.8337.48 ± 8.9034.45 ± 8.2319.950<0.001015C2 > C1***; C3 > C1*** TD 87.87 ± 19.66110.84 ± 23.83101.31 ± 22.5019.628<0.0010.15C2 > C1***; C2 > C3*;C3 > C1***Post hoc pairwise comparisons were performed using the Bonferroni correction. *P<0.05, **P<0.01, ***P<0.001 Comparison of SDQ and PSI-SF dimension results among different clustering groups of autistic children C2 > C1***; C2 > C3**; C3 > C1** C2 > C1**; C3 > C1* C2 > C1***; C3 > C1*** C2 > C1***; C3 > C1*** C2 > C1***; C3 > C1** C2 > C1***; C3 > C1*** C2 > C1***; C3 > C1*** C2 > C1***; C2 > C3*; C3 > C1*** Post hoc pairwise comparisons were performed using the Bonferroni correction. *P<0.05, **P<0.01, ***P<0.001 The impact of sensory characteristics on emotional and behavioral manifestations and parenting stress in children with ASD across different subgroups The fit indices of the structural equation models constructed separately for each of the three sensory subgroups of children with ASD were as follows: Model 1 (χ2/df = 1.20, P > 0.05; RMSEA = 0.03 90% CI [0 ~ 0.077], CFI = 0.995, SRMR = 0.021, TLI = 0.989); Model 2 (χ2/df = 1.21, P > 0.05; RMSEA = 0.03 90% CI [0 ~ 0.078], CFI = 0.995, SRMR = 0.025, TLI = 0.988); Model 3 (χ2/df = 1.43, P > 0.05; RMSEA = 0.04 90% CI [0 ~ 0.078], CFI = 0.986, SRMR = 0.028, TLI = 0.974). All three model fit indices met statistically acceptable criteria, indicating that the model exhibited a good fit [63, 64] and could effectively explain the relationships among sensory features, emotional/behavioral manifestations, and parenting stress in children with ASD. The structural equation modeling results showed that in Group 1 autistic children, the associations between movement sensitivity, auditory filtering, low energy/weak, sex, age and the SDQ total difficulties score, PSI-SF total parenting stress score were not statistically significant (all P > 0.05; Fig. 6(a)). In Group 2 autistic children, movement sensitivity had a significant direct negative effect on the SDQ total difficulties score (β = -0.26, P < 0.05), while auditory filtering had significant direct negative effects on both the SDQ total difficulties score and PSI-SF total parenting stress score (β = -0.25, -0.38, P < 0.05). The associations between other pathways were not statistically significant (P > 0.05; Fig. 6(b)). In Group 3 autistic children, auditory filtering had a significant direct negative effect on the PSI-SF total parenting stress score (β = -0.25, P < 0.01), and also exerted a significant indirect negative effect on the PSI-SF total parenting stress score through the mediation of the SDQ total difficulties score (β = -0.22, 0.31, P < 0.05). Additionally, child sex had a significant direct effect on the total parenting stress scores in Group 3. Caregivers of male children reported significantly higher stress than those of female children (β = 0.17, P < 0.05), while no statistically significant associations were observed among the remaining pathways (P > 0.05; Fig. 6(c)). Structural equation model of the effects of movement sensitivity, auditory filtering, low energy/weak, sex and age on SDQ total difficulties score and PSI-SF total parenting stress score across three autistic child clusters. (a) Cluster 1: χ2/df = 1.20, P > 0.05; RMSEA = 0.03 90% CI [0 ~ 0.077], CFI = 0.995, SRMR = 0.021, TLI = 0.989. (b) Cluster 2: χ2/df = 1.21, P > 0.05; RMSEA = 0.03 90% CI [0 ~ 0.078], CFI = 0.995, SRMR = 0.025, TLI = 0.988. (c) Cluster 3: χ2/df = 1.43, P > 0.05; RMSEA = 0.04 90% CI [0 ~ 0.078], CFI = 0.986, SRMR = 0.028, TLI = 0.974 Discussion Research summary This case-control study revealed that children with ASD exhibited significantly higher rates of sensory differences, emotional and behavioral manifestations, and parenting stress compared to the comparison group. Sensory processing issues were prevalent among autistic children, with the highest variability observed in three specific dimensions: movement sensitivity, auditory filtering, and low energy/weak. A k-means cluster analysis was conducted on the three SSP dimensions with the highest variability in children with ASD, dividing them into three sensory-based subgroups. Comparative analysis revealed significant differences in sensory profiles, emotional and behavioral manifestations, and parenting stress among the subgroups. Furthermore, the pathways through which sensory characteristics influenced emotional/behavioral manifestations and parenting stress varied across subgroups. Compared to the “sensory adaptive type” and “motor-sensitive/weakness type” subgroups, the “auditory filtering type” subgroup not only exerted a significant direct negative impact on parenting stress but also produced a significant indirect negative effect through the mediating role of emotional and behavioral manifestations. Comparison of sensory differences, emotional-behavioral manifestations, and parenting stress between the two groups of children Both clinical practice and research have found that children with ASD often exhibit significant sensory differences and behavioral manifestations, such as auditory filtering/sensitivity manifestations [65], selective eating behaviors due to tactile/taste differences [66], hyper-reactivity to other sensory inputs [67], and atypical behavioral responses to sensory stimuli [68]. This study compared 225 children with ASD and 228 comparison group participants, revealing that children with ASD showed significantly higher levels of sensory differences, emotional and behavioral manifestations, and parenting stress than the control group. Sensory differences were found to be prevalent among children with ASD. This is consistent with findings from other studies. Asaridou M et al. (2022) investigated sensory characteristics in 130 autistic children aged 8–12 years and 137 typically developing controls, revealing that sensory differences are prevalent among autistic children and show no sex differences [69]. Similarly, Riquelme et al. (2023) compared 38 autistic children with 34 typically developing controls and found that autistic children exhibited more pronounced sensory differences and behavioral manifestations than the TD group [10]. Additionally, Krakovich et al. (2016) demonstrated that parents of autistic children experience higher parenting stress compared to parents of typically developing children and those with other neurodevelopmental disorders [70]. The above research findings indicate that sensory differences, emotional and behavioral manifestations in children with ASD, as well as high levels of parenting stress, are prevalent issues that require greater attention and emphasis from both the nation and society. K-Means cluster analysis of sensory characteristics in children with ASD The k-means cluster analysis combined with the elbow method suggested that the optimal number of clusters in this study was three. Among them, Group 1 (33.33%) exhibited relatively high standardized mean values across all three dimensions and was labeled as the “sensory adaptive type”, Group 2 (29.78%) showed lower mean values in both the movement sensitivity and low energy/weak dimensions among autistic children and was designated as the “motor-sensitive/weakness type”, Group 3 (36.89%) demonstrated lower mean values in the auditory filtering dimension among autistic children and was classified as the “auditory filtering type”. Each subgroup displayed significant intergroup and intragroup differences across the three dimensions of movement sensitivity, auditory filtering, and low energy/weak. These three subgroups not only exhibited a high prevalence of sensory differences in autistic children but could also be distinctly differentiated based on their sensory profiles. This study classified the three most common sensory differences dimensions in children with ASD into characteristic subgroups, which can better categorize autistic children and guide the development of personalized intervention plans, demonstrating significant discriminative value in clinical diagnosis and early intervention. To identify distinct subgroups within ASD by utilizing sensory features, previous studies commonly performed K-Means clustering analysis on either the 38 items or 7 domains of the SSP. For instance, Kadlaskar et al. (2023) conducted cluster analysis on the 7 SSP domains of 211 autistic children, categorizing them into 4 groups with different sensory profiles [71]. Lyons-Warren et al. (2022) performed clustering analysis on the 38 SSP items and high/low sensitivity subsets of 378 autistic children, with consensus clustering-generated heatmaps indicating K = 3 as the most consistent clustering result [72]. Lane et al. (2014) clustered the 7 SSP domains of 228 autistic children into 4 groups with sensory feature differences, analyzing associations between subgroups and ASD core symptoms/nonverbal IQ [73]. Shirley et al. (2024) utilized sensory features, behavioral characteristics, and perinatal risk factors from Australia’s ASD sample registry (n = 1168) for cluster analysis, identifying 4 clinically meaningful subgroups to inform intervention and prognosis prediction [74]. However, current research on cluster analysis of sensory characteristics in children with ASD remains relatively scarce overall, with most studies focusing on all items or all dimensions of the SSP, lacking clinical guidance. Therefore, this study conducts K-Means cluster analysis on the three dimensions with the highest prevalence of sensory differences in children with ASD. This cluster-based approach allows for clearer characterization of ASD clinical presentations and sensory subtypes, providing a more targeted foundation for exploring early personalized interventions for autistic children with distinct sensory features. The association between sensory characteristics of children with ASD and their emotional-behavioral manifestations and parenting stress This study found that more pronounced sensory differences in children with ASD are often accompanied by heightened emotional and behavioral difficulties and lower levels of prosocial behavior. The degree of sensory differences is positively correlated with parental parenting stress. Meanwhile, the level of children’s emotional and behavioral difficulties is positively correlated with parental parenting stress, whereas the level of prosocial behavior is negatively correlated with family parenting stress. As a core symptom of ASD, sensory differences are associated with clinical symptoms and prognosis in children. Those with more pronounced sensory differences tend to exhibit more emotional and behavioral manifestations, and their parents experience higher levels of parenting stress. Other researchers have also supported this view. Riccio et al. (2025) found in their study of 52 children with ASD that the degree of sensory differences was correlated with behavioral manifestations [66]. Similarly, Ausderau et al. (2016) demonstrated a correlation between sensory differences in children with ASD and parenting stress [41]. A comparison of emotional and behavioral manifestations in children with ASD and parenting stress across different subgroups revealed that Group 2 (motor-sensitive/weakness type) scored the highest, exceeding both Group 3 (auditory filtering type) and Group 1 (sensory adaptive type). Analysis of the three subgroups showed that Group 2 was characterized by lower mean values in movement sensitivity and low energy/weak features, with auditory filtering features also scoring relatively low. This indicates that among the three subgroups, Group 2 demonstrated the most pronounced sensory features across all three most common sensory differences dimensions in SSP, representing the most severe sensory differences. Group 3 ranked second, showing positive standardized mean values in movement sensitivity and low energy/weak dimensions but negative values in auditory filtering. Moreover, Group 3 showed fewer differences in the auditory filtering dimension compared with Group 2. Finally, Group 1 exhibited the mildest sensory differences across all three dimensions. The results above indicate that there is an association between the magnitude of sensory differences and emotional and behavioral manifestations in children with ASD as well as parenting stress across the three subgroups. The subgroup with more pronounced sensory differences exhibited more pronounced emotional and behavioral manifestations and higher levels of parenting stress. According to the SSP assessment content [46], children with ASD who exhibit auditory filtering differences often demonstrate poor response to their names, difficulty concentrating, and distractibility. Those with movement sensitivity typically show heightened reactions to motion stimuli, such as disliking being held or fearing heights. Children with low energy/weak frequently present with easy fatigability, limited daily activities, and increased risk of accidents. This indicates that sensory differences in these three domains can severely impact the daily life and learning of children with ASD, significantly elevate emotional and behavioral manifestations, and exacerbate parental caregiving stress. Structural equation modeling of the impact of sensory characteristics in children with ASD on emotional-behavioral manifestations and parenting stress The structural equation modeling results indicated no significant associations between movement sensitivity, auditory filtering, low energy/weak and the SDQ total difficulties score or PSI-SF total parenting stress score in Group 1 autistic children. The findings demonstrate that Group 1 autistic children exhibited the mildest sensory differences across these three sensory dimensions. Therefore, among autistic children with less pronounced sensory differences, sensory features do not exert significant influences on emotional and behavioral manifestations or parenting stress. The movement sensitivity of autistic children in Group 2 had a significant direct negative effect on the SDQ total difficulties score (β=-0.26), while auditory filtering showed significant direct negative effects on both the SDQ total difficulties score and the PSI-SF total parenting stress score (β=-0.25, -0.38). As shown in Fig. 5, Group 2 exhibited the lowest standardized mean values across all three sensory dimensions relative to the other two clusters, indicating the most pronounced sensory differences in these domains. However, only auditory filtering demonstrated a direct significant impact on parenting stress, with no indirect significant effects observed for auditory filtering, movement sensitivity or low energy/weak on the PSI-SF total score through the mediating role of SDQ. The auditory filtering of children with ASD in Group 3 could exert a significant negative direct effect on the total PSI-SF parenting stress score (β=-0.25), and could also indirectly produce a significant negative effect on the total PSI-SF parenting stress score through the mediation of SDQ total difficulties score (β=-0.22, 0.31). Additionally, sex showed a significant positive effect on the total PSI-SF parenting stress score, Caregivers of male children reported significantly higher stress than those of female children (β = 0.17). The standardized mean values of Group 3 ASD child in the movement sensitivity and low energy/weak dimensions were positive, with only mild auditory filtering differences. Moreover, the differences in these three dimensions were significantly lower than those in Group 2. This indicates that subgroups composed of different sensory profiles have significantly varying impacts on the emotional and behavioral manifestations of autistic children and parenting stress. autistic children characterized by auditory filtering differences can significantly influence parenting stress through both direct effects and indirect effects mediated by SDQ. Previous studies have shown no significant sex differences in the impact on parenting stress of autistic children [75, 76]. The findings of this study specifically pertain to “auditory filtering type” autistic children, and conclusions regarding sex differences require further validation. A series of recent cutting-edge studies have employed structural equation modeling to conduct detailed investigations into parenting stress, family impact, and related influencing factors among parents of children with ASD. For instance, Salami et al. (2025) revealed through structural equation modeling that the clinical symptoms of children with ASD could significantly impact parental quality of life through family coping strategies and family functioning [75]. Similarly, Chen et al. (2025) demonstrated via structural equation modeling that caregiver impact in parents of children with ASD could significantly influence family resilience through social support and illness cognition [77]. Additionally, Rattaz et al. (2025) used structural equation modeling to show that oppositional behaviors in children with ASD significantly affected parental quality of life through parenting stress and coping styles [78]. Previous studies have employed structural equation modeling to explore the pathways through which clinical symptoms and oppositional behaviors in children with ASD indirectly affect parental quality of life. This study innovatively demonstrates that identifying distinct subgroups of autistic children through sensory features offers a promising approach for recognizing ASD phenotypes. It investigates the behavioral foundations closely associated with parenting stress in autistic children’s parents and examines the pathways by which different sensory features in autistic children influence their emotional and behavioral manifestations as well as parental stress. This study demonstrates how cluster analysis of sensory profiles can reveal distinct subgroups of autistic children with clinically meaningful characteristics. Specifically, we identified an auditory filtering subtype, which exerts a dual negative impact on parenting stress. The findings suggest that clinical practice should prioritize autistic children with auditory filtering difficulties, where structured environmental supports to reduce auditory load may help improve children’s emotional and behavioral functioning, thereby alleviating caregiver distress. Beyond individual-level interventions, these results also highlight the need for policy-level support, such as targeted mental health resources for caregivers of children with sensory-related challenges, and provide a theoretical basis for developing more targeted, multi-level early intervention strategies based on sensory subtypes. Conclusions Children with ASD exhibited significantly higher rates of sensory differences, emotional and behavioral manifestations, and parenting stress compared to the control group. Sensory differences were prevalent among autistic children, with three dimensions being most prominent (all > 50%): movement sensitivity, auditory filtering, and low energy/weak. Using the K-Means clustering method combined with the elbow criterion, the three dimensions with the highest sensory discrepancy rates in autistic children (movement sensitivity, auditory filtering, and low energy/weak) were categorized into three subgroups. The distinct subgroups identified in this study were characterized as “sensory adaptive type”, " motor-sensitive/weakness type” and “auditory filtering type”. A correlation existed between sensory differences in autistic children and their emotional/behavioral manifestations as well as parenting stress—autistic children with more pronounced sensory differences were more likely to have greater emotional and behavioral manifestations and elevated parenting stress. In the “auditory filtering type” autistic children, sensory differences could influence parenting stress through both direct effects and indirect effects mediated by the SDQ. This study suggests that classification based on sensory differences provides a promising approach for identifying ASD phenotypes. Sensory-based phenotypic characteristics may help identify behavioral traits that respond better to specific interventions, thereby improving emotional and behavioral manifestations in children with ASD, alleviating parenting stress. Limitations of the study This study is a cross-sectional survey, which can only indicate correlations between variables and cannot establish causal relationships. Additionally, all questionnaires involved were completed by the parents of the children, inevitably introducing potential parental reporting bias. Future research should consider longitudinal studies and incorporate reports from teachers and clinicians to obtain more accurate conclusions. This study lacks assessments of the intelligence quotient (IQ) and social functioning of children with autism, and comprehensive developmental evaluations should be included in subsequent research. Furthermore, all samples in this study were from Nantong City, which imposes certain limitations on the generalizability of the findings. Future studies should aim to expand the sample scope to achieve more representative results. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Author contributions ZX: Writing – original draft, Writing – review & editing. FW: Writing – review & editing. HJ, LM, QZ, YX: Investigation, Writing – review & editing., LW, JW: Conceptualization, Investigation, Writing – review & editing. Funding This study was supported by the National Natural Science Foundation of China (U24A20770). Data availability The survey data supporting the findings of this study are available from the School of Public Health, Harbin Medical University, but restrictions apply to the availability of these data, which were used under license for the current study and so are not publicly available. However, the data may be obtained from the School of Public Health, Harbin Medical University upon reasonable request and with permission. Declarations Ethical approval and consent to participate This study was conducted in accordance with the ethical guidelines outlined in the Declaration of Helsinki and was approved by the Ethics Committee of Nantong Maternal and Child Health Hospital (Ethics Approval No. Y2024002). Prior to the study, informed consent was obtained from the primary guardians of children with ASD, and all participating families signed informed consent forms. All data were processed anonymously. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Zhanbin Xu and Feiying Wang contributed equally to this work. Contributor Information Jia Wang, Email: wangjiahyd@163.com. Lijie Wu, Email: wulijiehyd@126.com. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The survey data supporting the findings of this study are available from the School of Public Health, Harbin Medical University, but restrictions apply to the availability of these data, which were used under license for the current study and so are not publicly available. However, the data may be obtained from the School of Public Health, Harbin Medical University upon reasonable request and with permission.