The effect of fully immersive virtual reality technology combined with psychological and behavioral intervention on autism spectrum disorder

Abstract

Objective

At present, there is a lack of targeted support models for individuals with autism spectrum disorder (ASD), mainly relying on comprehensive treatment with rehabilitation, education, and medication as supplements. Although there is some effectiveness, there are shortcomings such as long treatment cycles and limited engagement from some autistic children, especially in traditional formats. This study explores the effectiveness of fully immersive virtual reality (FIVR) combined with psychological and behavioral interventions for ASD.

Methods

In this retrospective cohort study, 124 children with autism who received treatment at Hongxinkang New Traditional Chinese Medicine Hospital in Tongren from January 2024 to July 2024 were included. 62 patients who received FIVR combined with psychological and behavioral intervention were matched in a 1:1 ratio with the queue receiving psychological and behavioral intervention. The main endpoint is the Childhood Autism Rating Scale (CARS) and Aberrant Behavior Checklist (ABC), as well as the scores of Psychoeducational Profile-third edition (PEP-3) after three months of intervention. The secondary outcome is the satisfaction of the child’s family members.

Results

After three months of intervention, both groups showed reductions in ABC and CARS scores, with the FIVR group exhibiting more pronounced improvements. ANCOVA confirmed significant adjusted group differences favoring the FIVR group (ABC adjusted mean difference = − 5.67, 95% CI [–6.34, − 5.01], partial η² = 0.712; CARS adjusted mean difference = − 3.36, 95% CI [–4.10, − 2.61], partial η² = 0.408). Similarly, PEP-3 total scores were significantly higher in the FIVR group (adjusted mean difference = 8.21, 95% CI [6.48, 9.95], partial η² = 0.430), with consistent gains across subdomains, particularly in language and adaptive behavior. Family satisfaction was also greater in the FIVR group (95.2% vs. 82.3%; χ² = 5.153, P = 0.023; Cramér’s V = 0.20, 95% CI [0.012, 0.362]).

Conclusions

Integrating FIVR with psychological and behavioral interventions may yield robust improvements in behavioral regulation, autism severity, neuropsychological development, and caregiver satisfaction. By incorporating ANCOVA adjusted for baseline covariates and reporting effect sizes with 95% confidence intervals, the robustness and clinical relevance of these findings were strengthened. Nevertheless, given the retrospective and non-randomized design, these results should be interpreted as preliminary. Future prospective randomized controlled trials are essential to confirm the effectiveness, durability, and mechanisms of FIVR-based interventions.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40359-025-03460-y.

Introduction

Autism spectrum disorder (ASD) is a neurodevelopmental condition characterized by persistent deficits in social communication and social interaction across multiple contexts, alongside restricted, repetitive patterns of behavior, interests, or activities. According to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, Text Revision (DSM-5-TR, 2022), these symptoms typically appear in early developmental stages and lead to clinically significant impairment in social, academic, occupational, or other areas of functioning [1, 2]. ASD typically manifests during early childhood, with a higher prevalence observed in males [2, 3]. Research shows that the incidence rate of ASD is increasing year by year in recent years [1–3]. Among them, many autistic individuals may face lifelong challenges in achieving independent living, which places significant demands on families and support systems [1–4]. Therefore, early intervention for ASD has a positive significance in improving the physical and mental health of children with ASD, and reducing the burden on families and society [3, 4]. Unlike previous studies that applied virtual reality (VR) as a stand-alone tool or within non-immersive settings, our study featured a protocolized fully immersive virtual reality (FIVR) combined with psychological and behavioral interventions (PBI) model, tailored intervention scenes, and evaluation in a culturally specific clinical setting. These elements collectively expand current knowledge and offer a feasible model for integration of immersive technologies in ASD rehabilitation within non-Western contexts.

In recent years, there has been increasing interest in applying VR technologies to support children with ASD, ranging from desktop-based simulations to semi- and fully immersive systems. For instance, Karami et al. [5] conducted a meta-analysis showing that VR-based interventions moderately improve emotional recognition and social functioning in ASD. However, most studies employed semi-immersive platforms or limited virtual interaction. Dechsling et al. [6] reviewed 20 VR/AR-based social skills training studies and highlighted the lack of standardized protocols and difficulty in generalizing outcomes. Yuan et al. [7] demonstrated that immersive VR can enhance emotional expression and social motivation but noted challenges in designing scenarios adapted to individual cognitive needs. Despite promising findings, current VR-based interventions face several limitations. These include insufficient personalization, low ecological validity, and limited adaptability to diverse cultural contexts. Moreover, many programs lack integration with evidence-based psychological-behavioral frameworks, which are essential for sustained therapeutic impact. While psychological and behavioral interventions (PBI) are well-established in ASD rehabilitation, their integration with FIVR has been insufficiently investigated in clinical practice—particularly within non-Western populations [3, 5]. While prior studies have investigated the use of virtual reality (VR) in ASD interventions, our study provides several distinctive contributions. First, we systematically integrated fully immersive VR (FIVR) with a standardized psychological and behavioral intervention (PBI) protocol, and implemented it in a clinical rehabilitation setting in China—a context that has been underrepresented in the literature. Second, the FIVR modules were individually tailored based on initial assessments of sensory sensitivity, social anxiety, and language impairments. Third, we employed multidimensional outcome measures, including not only CARS, ABC, and PEP-3, but also caregiver satisfaction, offering a holistic evaluation of intervention effectiveness from both clinical and family-centered perspectives.

The rationale for using fully immersive virtual reality (FIVR) in ASD interventions can be supported by several theoretical frameworks. Social motivation theory suggests that autistic individuals often experience diminished motivation for social engagement [7]. FIVR enables controlled, low-pressure, and repetitive social scenarios, thereby reducing anxiety and gradually enhancing motivation and competence in social communication. Sensory integration theory posits that atypical sensory processing—frequently observed in autistic children—can be mitigated through graded sensory exposure [8]. FIVR provides a customizable environment where visual, auditory, and tactile inputs can be modulated to support sensory adaptation and regulation. Additionally, the executive function scaffolding model emphasizes that task-based VR environments can enhance planning, inhibitory control, and cognitive flexibility, which are often underdeveloped in ASD populations [9]. These mechanisms collectively support the use of FIVR as a multisensory and interactive training platform tailored to the neurodevelopmental needs of autistic children.

At present, there is a lack of targeted support models for individuals with ASD, mainly relying on comprehensive treatment with rehabilitation, education, and medication as supplements [1, 10, 11]. Although there is some effectiveness, there are shortcomings such as long treatment cycles and limited engagement from some autistic children, especially in traditional formats [10–12]. The existing conventional language cognitive training, sensory integration training, group lesson mode training, transcranial magnetic therapy, game training, traditional Chinese medicine treatment and other combination modes for parents and children have become exhausted [11, 12]. The concept of psychological and behavioral intervention has been applied in the field of ASD for many years in developed regions such as the United States, Taiwan, and Hong Kong [13, 14]. However, the application of ASD rehabilitation and education in China is seriously lagging behind. While existing studies have demonstrated the potential of VR-based interventions for children with ASD, the majority employ non-immersive or semi-immersive systems, and often lack standardization in therapeutic protocols [15, 16]. Furthermore, few studies have explored the clinical integration of fully immersive virtual reality (FIVR) with structured psychological and behavioral interventions (PBI), especially in non-Western populations. Given cultural, educational, and infrastructural differences, there is a pressing need to assess the feasibility and effectiveness of FIVR-based PBI in the Chinese context. To address this gap, the present study combines standardized PBI with fully immersive VR technology in a structured intervention model. We aim to evaluate its impact on behavioral symptoms and developmental domains in children with ASD, thereby providing preliminary clinical evidence for the integration of immersive digital tools into conventional autism therapy models in China.

Therefore, the objective of this study is to evaluate the effectiveness of combining FIVR with PBI in improving outcomes among children with autism spectrum disorder (ASD). Specifically, we examine its impact on symptom severity, behavioral profiles, neuropsychological development (including personal-social, language, and motor domains), and caregiver satisfaction.

Materials and methods

Research design and patients

This is a single center retrospective cohort study conducted at Hongxinkang New Traditional Chinese Medicine Hospital in Tongren. We included 124 children with ASD who received treatment from January 2024 to October 2024. A total of 62 pediatric patients received FIVR combined with psychological and behavioral interventions. A matched cohort of 62 pediatric patients who received pure psychological and behavioral intervention was selected using a 1:1 manual matching approach. Matching was based on four key baseline characteristics: (1) age (± 1 year), (2) gender, (3) duration of diagnosis (disease duration), and (4) initial severity of ASD, as assessed by baseline CARS and ABC scores. The matching was performed by trained clinical staff blinded to post-intervention outcomes during retrospective chart review. Propensity score matching (PSM) was not applied due to the relatively small sample size, limited covariates, and the single-center nature of the dataset. All participants included in the analysis were successfully matched, and no cases were excluded due to matching failure. All included participants completed the full three-month intervention protocol and outcome assessments. No dropouts or losses to follow-up occurred during the study period. Baseline comparability between the two groups was confirmed by statistical testing (see Table 1), with no significant differences across matched variables. All procedures performed in study involving human participants were in accordance with the ethical standards of the institutional and/or national research committee(s) and with the Helsinki Declaration (as revised in 2013). The informed consent was waived by the Ethics Review Board of Tongren Hongxin Kangxin Traditional Chinese Medicine Hospital for the observational and retrospective nature.

Table 1.

Comparison of general information between two groups

General Information	FIVR group (n = 62)	Control group (n = 62)	t/χ2	P	Cramér’s V/ Rank-biserial r	95% CI
Male (yes), n (%)	35 (56.5)	40 (64.5)	0.844	0.358	< 0.001		(-0.177, 0.177)
Age (years), M(P25/P75)	6 (5–9)	8 (5–9)		0.340*	0.098		(-0.096, 0.297)
Disease duration (years), M(P25/P75)	2.5 (2–4)	2 (2–3)		0.143*	-0.137		(-0.312, 0.039)
Primary guardian, n (%)
Father	17 (27.4)	10 (16.1)	4.795	0.091
Mother	28 (45.2)	40 (64.5)			0.150		(-0.025, 0.326)
Ancestors	17 (27.4)	12 (19.4)
Educational background of primary guardian, n (%)
Junior high school and below	36 (58.1)	38 (61.3)	0.134	0.714	< 0.001		(-0.177, 0.177)
High school and above	26 (41.9)	24 (38.7)
Single parent (yes), n(%)	21 (33.9)	19 (30.6)	0.148	0.701	< 0.001		(-0.177, 0.177)

Note: ^* P values derived from Mann–Whitney U test

Inclusion criteria: (1) Diagnosis of autism spectrum disorder (ASD) based on the DSM-5 diagnostic criteria, as assessed by licensed pediatric neurologists and developmental specialists through structured clinical interviews, behavioral observations, and developmental history reviews. Standardized diagnostic tools such as the Autism Diagnostic Observation Schedule (ADOS) or the Social Communication Questionnaire (SCQ) were not consistently administered due to clinical resource constraints at the time of data collection; [1] (2) Patients aged 3–12 with stable and conscious consciousness; (3) Complete clinical data. Exclusion criteria: (1) Patients with concomitant Asperger’s syndrome; (2) Individuals with combined schizophrenia; (3) Patients with delayed mental development; (4) Patients with combined epilepsy; (5) Patients with visual and auditory impairments; (6) Patients with a history of intracranial surgery or implantation of metal foreign bodies.

Intervention methods

Psychological and behavioral intervention

(1) Language practice. Inducing children to enhance communication through understanding language exercises, strengthening pronunciation, attention exercises, etc.; two times/week, 30 min/time. (2) Personal sensory integration practice. Adopting balance perception exercises, tactile exercises, and proprioceptive exercises to cultivate sensory integration abilities; three times/week, 30 min/time. (3) Group practice. Group-based activities (e.g., jigsaw puzzles, toy assembly, bead stringing, dancing) were conducted in sessions of 4 to 6 children, grouped based on developmental level and behavioral compatibility; two times/week, 30 min/time. (4) Personalized exercises. Design courses such as sports, games, painting, music, etc. based on the individual interests and preferences of the affected child; two times/week, 30 min/time. (5) Family practice. Invite the families of children with autism to actively participate in structured recreational activities; Accompany and guide children to actively participate in outdoor activities, enhance communication and emotional connections between children and their families. (6) Psychological intervention. Enhance communication with pediatric patients, maintain a warm, friendly, and supportive demeanor during therapist-child interaction, and create a warm and comfortable environment; And during communication, look into the eyes of the child and observe their expressions and changes in their gaze; Provide timely emotional support, psychological counseling, and comfort.

FIVR

All participants received a structured intervention program for a total duration of 12 weeks (3 months). The psychological and behavioral intervention (PBI) was conducted six sessions per week, while FIVR training was delivered 3 to 5 times weekly, depending on individual adaptability and availability.

(1) Evaluation phase. Initial assessment- Conduct a comprehensive evaluation of children with ASD, including personal-social (such as eye contact, perception of others’ presence, frequency of social interaction, etc.), communication skills (language expression, understanding instructions, etc.), repetitive stereotyped behaviors (such as frequency and intensity of clapping, shaking the body, etc.), and sensory sensitivity (response to sound, light, touch, etc.). Personalized needs analysis- Based on the evaluation results, analyze the child’s individual functional profile (e.g., sensory sensitivities, level of verbal communication, adaptive behavior). Based on these assessments, FIVR scenes were partially individualized, allowing customized selection of training modules to target specific developmental goals within the standardized VR framework; If the child is sensitive to visual stimuli, special attention should be paid to the selection and presentation intensity of visual elements when designing VR scenes. (2) VR scene design. Social scenarios- Create virtual social environments, such as virtual schools, virtual parks, etc.; Arrange virtual characters (with different appearances, genders, and age characteristics) to interact with children with autism in the virtual school scene; Virtual classmates can proactively greet children with autism and invite them to participate in activities (such as playing games, attending classes, etc.). Design multi person interactive plots in virtual park scenes; Gradually guide the children to participate in social interactions, such as flying kites or having picnics together; Improve their understanding and adaptive behavior to social situations; Communication scenario- Build a virtual scenario specifically designed for language and nonverbal communication training; In a virtual store scenario, the child needs to communicate with a virtual salesperson to purchase items; This includes expressing one’s own needs (such as the name and quantity of the item) and understanding the salesperson’s answers (such as price, availability, etc.). Non verbal communication scenarios can also be set up; Conveying emotions and intentions through facial expressions and body movements; At a virtual birthday party, the child needs to respond appropriately based on the expressions of others (happy, surprised, etc.) and body movements (hugging, waving, etc.). Sensory regulation scene- Create a virtual scene for sensory regulation for children with ASD who are sensitive to their senses; For children with auditory sensitivity, a quiet virtual forest environment with soft ambient sounds (e.g., breeze, birdsong) can be designed to facilitate gradual sensory adaptation. Design virtual tactile experience scenes for children with tactile sensitivity; If the child is allowed to touch different objects (soft feathers, rough bark, etc.) in a virtual environment to adjust their tactile sensitivity. (3) Rehabilitation training plan. Training frequency: 3–5 VR rehabilitation training sessions per week; The duration of each training session depends on the child’s attention and tolerance level; The initial stage can be set to 20–30 min, gradually extending to 40–60 min as the child’s adaptive behavior improves. Training steps- Warm up phase (5–10 min) - Before entering a specific rehabilitation scenario, allow the child to adapt to VR devices in a simple and relaxed virtual environment (such as a virtual children’s room); Observe the emotions and reactions of the child and engage in simple interactions with them; Such as selecting toys in the room through joystick operation. Rehabilitation training stage (10–40 min) - Based on the rehabilitation needs of the child, enter the corresponding scene for training. Relaxation stage (5–10 min) - After completing rehabilitation training, allow the child to return to a relaxed virtual environment; Like a virtual seaside, allowing children to stroll and watch the waves on the beach helps them relax and relieve stress during training. (4) Data monitoring and adjustment. Data collection- Collect various data of pediatric patients during VR rehabilitation training; Such as the number of social interactions, accuracy of language expression, frequency of repetitive stereotyped behaviors, heart rate, skin conductance response, etc. Assess the physiological and psychological state of the child during the training process. Plan adjustment- Analyze and adjust the rehabilitation plan on a weekly basis based on the collected data; If the child’s progress is not significant in a certain scene, or the number of interactions in the virtual social scene does not increase, the difficulty of the scene can be adjusted (such as increasing the number of virtual characters or changing the interaction mode) or the training method can be adjusted (such as increasing the frequency of prompts or changing the prompt mode).

Collect data

Collect baseline patient data, including gender, age, disease duration, and family background. Collect relevant evaluation indicators for patients before and 3 months after intervention: (1) behavioral status. According to the Autism Behavior Checklist (ABC) assessment [17]. It is completed by the primary nursing staff. The scale includes five aspects: sensation, relationship, stereotyped behavior, language, and social independence, with a critical value of 53 for ASD. (2) Severity of illness. The Childhood Autism Rating Scale (CARS) is used by developmental pediatricians to assess the severity of ASD and consists of 15 items [18]. Each project is rated on a continuum from normal to severely abnormal. A score of one indicates a normal range for age, two indicates a mild abnormality, three indicates a moderate abnormality, and four indicates a severe abnormality. The total score ranges from 15 to 60, with scores of 30–36 indicating mild ASD and 36 or more indicating severe ASD. (3) Neuropsychological function. According to the pediatric neuropsychological development assessment checklist, including personal-social, language, adaptive behavior, fine motor skills, and gross motor skills, higher scores indicated better functional performance; Using the developmental quotient as an indicator, < 70 points are considered developmental disorders, 70–79 points are considered critically low, 80–109 points are considered moderate, 110–130 points are considered good, and > 130 points are considered excellent. (4) Ability development level. According to the Psychoeducational Profile-third edition (PEP-3) assessment; [19] Including sensory response, imitation, cognitive expression, emotional expression, hand eye coordination, etc., a total of 166 points; The higher the score, the better the development of abilities. (5) Family satisfaction was investigated using the Newcastle Satisfaction with Nursing Scales (NSNS) after nursing care. The scale consisted of 19 items, with a total score of 95 points (five points for each item, three points for general satisfaction, and one point for dissatisfaction); Among them, a score of ≥ 76 indicates satisfaction, 56–75 indicates general satisfaction, and 19–55 indicates dissatisfaction. The number of patients with different levels of satisfaction will be counted; Nursing satisfaction=(number of satisfied cases + number of generally satisfied cases) / total cases × 100%.

The psychometric properties of the assessment tools used in this study have been well-established. The Aberrant Behavior Checklist (ABC) has demonstrated high internal consistency across subscales, with Cronbach’s α ranging from 0.86 to 0.94 in ASD populations. The Childhood Autism Rating Scale (CARS) exhibits excellent inter-rater reliability (r = 0.88–0.94) and internal consistency (Cronbach’s α ≈ 0.94), and its validity has been confirmed in Chinese clinical samples. The Psychoeducational Profile – Third Edition (PEP-3) shows robust psychometric characteristics, including internal consistency coefficients ranging from 0.82 to 0.96 and strong construct validity, as validated in simplified Chinese versions [19].

Statistical analysis

All data were input into Microsoft Excel and analyzed using SPSS version 26.0 (IBM Corp, Armonk, NY, USA). The Shapiro–Wilk test was applied to assess the normality of continuous variables. Given the non-normal distribution of most variables, continuous data (e.g., ABC, CARS, neurodevelopmental scores, and PEP-3 scores) were expressed as median and interquartile range. For between-group comparisons, the Mann–Whitney U test was used; for within-group comparisons (pre- vs. post-intervention), the Wilcoxon signed-rank test was applied. Categorical variables (e.g., gender, family satisfaction levels) were analyzed using chi-square tests. To account for potential baseline differences, analysis of covariance (ANCOVA) was conducted for the primary outcomes (ABC, CARS, and PEP-3), adjusting for covariates such as age, sex, diagnosis duration, and baseline severity. Effect sizes (Rank-biserial r or Cramér’s V for categorical data, and partial η² for ANCOVA models) were calculated and reported with 95% confidence intervals. Model assumptions were examined by QQ-plots and residual histograms, which are provided in Supplementary Figures S1–S6. Subdomain analyses of PEP-3 are summarized in Supplementary Table 1. GraphPad Prism version 8.0 (GraphPad Software, San Diego, USA) was used to visualize changes in outcome indicators over the three-month intervention period. A two-sided P-value < 0.05 was considered statistically significant.

Results

Baseline characteristics were comparable between the FIVR and control groups (Table 1). No statistically significant differences were observed in terms of sex, age, disease duration, or baseline scores on the ABC, CARS, and PEP-3 scales (all P > 0.05). Sociodemographic variables, including the identity of the primary guardian (father, mother, or grandparent), caregiver education level (junior high school and below vs. high school and above), and single-parent household status, also showed no significant group differences. These results suggest overall balance in clinical and family-related baseline parameters. Effect sizes were small, and their 95% confidence intervals all included the null, suggesting no clinically meaningful baseline imbalance between groups.

Severity of illness and behavioral status

As shown in Fig. 1, both groups experienced a reduction in ABC and CARS scores after the 3-month intervention period. However, the FIVR group demonstrated a more pronounced decrease in median scores compared to the control group. Consistent with these visual patterns, ANCOVA confirmed significant adjusted group differences (Table 2). For ABC, the adjusted mean difference was − 5.67 (95% CI [–6.34, − 5.01], P < 0.001, partial η² = 0.712), indicating substantial improvements in behavioral regulation. For CARS, the adjusted mean difference was − 3.36 (95% CI [–4.10, − 2.61], P < 0.001, partial η² = 0.408), suggesting robust alleviation of autism severity in the FIVR group compared with controls.

Fig. 1.

Comparison of Disease Severity and Behavioral Status between Two Groups; Aberrant Behavior Checklist (ABC); Childhood Autism Rating Scale (CARS)

Table 2.

Comparison of adjusted group differences in core outcomes via ANCOVA

Outcome	Adjusted Mean Difference	95% CI	P-value	Partial η²
ABC	-5.67	(-6.34, -5.01)	< 0.001	0.712
CARS	-3.36	(-4.10, -2.61)	< 0.001	0.408
PEP-3	8.21	(6.48, 9.95)	< 0.001	0.430

Neuropsychological function

As depicted in Fig. 2, both groups experienced post-intervention improvements in fine motor skills, language comprehension, and cognitive function (P < 0.05). However, the FIVR group showed significantly greater gains across these domains. ANCOVA confirmed that the FIVR group achieved significantly higher overall PEP-3 scores than the control group (adjusted mean difference = 8.21, 95% CI [6.48, 9.95], P < 0.001, partial η² = 0.430; Table 2). These findings suggest that immersive VR provides enriched and structured learning contexts that can enhance neurodevelopmental stimulation beyond conventional behavioral training.

Fig. 2.

Comparison of neurological and psychological functions between two groups

Ability development level

As illustrated in Fig. 3, both groups demonstrated post-intervention improvements in developmental abilities as assessed by the PEP-3. The FIVR group achieved significantly greater gains compared with the control group. ANCOVA confirmed that the FIVR group had higher adjusted PEP-3 total scores (adjusted mean difference = 8.21, 95% CI [6.48, 9.95], P < 0.001, partial η² = 0.430; Table 2). Subdomain analyses further demonstrated consistent improvements across all five domains (personal-social, language, adaptive behavior, fine motor, and gross motor). The largest effects were observed in language (adjusted mean difference = 7.56, 95% CI [6.83, 8.30], partial η² = 0.781) and adaptive behavior (adjusted mean difference = 7.33, 95% CI [6.56, 8.11], partial η² = 0.752) (Supplementary Table 1). These findings indicate that immersive VR provided enriched developmental stimulation and promoted significant improvements across multiple ability domains beyond conventional behavioral training.

Fig. 3.

Compares the developmental levels of two groups PEP-3 scores. Psychoeducational Profile-third edition (PEP-3)

Satisfaction of the families of the children with autism

Family satisfaction outcomes are summarized in Table 3. The proportion of caregivers who were “very satisfied” or “satisfied” was higher in the FIVR group (95.2%) than in the control group (82.3%). This difference was statistically significant (χ² = 5.153, P = 0.023). The effect size (Cramér’s V = 0.20, 95% CI [0.012, 0.362]) indicated a small-to-moderate association between intervention type and caregiver satisfaction. These findings suggest that immersive VR interventions were not only clinically effective but also well-accepted by families in terms of satisfaction.

Table 3.

Comparison of family satisfaction between two groups of ASD

Group	n	Very satisfied	Satisfied	Dissatisfied	Overall satisfaction
FIVR group	62	31 (50.0)	28 (45.2)	3 (4.8)	59 (95.2)
Control group	62	26 (42.0)	25 (40.3)	11 (17.7)	51 (82.3)
χ 2					5.153
P					0.023
Cramér’s V					0.200
95% CI					(0.012, 0.362)

Discussion

This study analyzed the effectiveness of FIVR combined with psychological and behavioral interventions for children with ASD. The results of this article confirm that the combined use of FIVR and psychological behavioral intervention is associated with improvements in behavioral symptoms, neuropsychological development, and increased caregiver satisfaction among children with ASD. Importantly, by supplementing the initial non-parametric analyses with ANCOVA adjusted for baseline covariates, we demonstrated that these group differences remained statistically robust. The inclusion of effect sizes and 95% confidence intervals provides a clearer picture of the magnitude and clinical relevance of the intervention effects. There is evidence to suggest that psychological and behavioral interventions are an important component of the treatment for children with ASD [13, 14]. It can help them improve their social interaction, communication skills, and self-management abilities [13, 14, 20]. However, psychological and behavioral intervention requires long-term investment and patience, which is a challenge for both parents and professionals [20, 21]. Moreover, the effectiveness of intervention varies from person to person and may also cause resistance among the affected children or their families [14, 20–22]. Compared with previous studies that solely focused on the impact of psychological and behavioral interventions on children with ASD, this study combined FIVR. This study comprehensively considers multiple dimensions, including disease severity, behavioral status, neuropsychological function, ability development, and family satisfaction. It comprehensively considers the effectiveness of intervention measures and can more accurately reflect the overall impact of intervention measures on children and their families.

The results of this study showed that after three months of intervention, the CARS and ABC scores of the FIVR group were lower than those of the control group. This is consistent with the findings of Koushki et al. [23] In addition, Papathomas et al. [24] showed that FIVR can provide richer and more realistic simulation environments for children, helping them better adapt to different social scenarios, behavioral requirements, etc.; Children can practice interacting with virtual characters in a relatively safe and controllable environment; These improvements in virtual environments may translate into enhanced personal-social functioning and fewer autism-related behavioral challenges in daily life. FIVR can create personalized environments tailored to the unique cognitive profiles, sensory preferences, and communication styles of autistic children [23, 24]. There are differences in the symptoms and severity of ASD among different children, and FIVR can design different training scenarios, task difficulty, etc. based on individual differences. This may improve the precision of intervention and be associated with enhanced rehabilitation outcomes [6, 14, 23]. Meanwhile, some scholars have also found that the immersion and fun of VR can attract the attention of children with ASD, who are usually more interested in novel and interesting things. Compared with traditional intervention methods, FIVR facilitated children’s active engagement in the intervention process, contrasting with the passive reception often observed in conventional training [6, 25].

The results of this study also showed that after three months of intervention, the FIVR group had significantly higher scores in PEP-3, personal-social, language, adaptive behavior, fine motor skills, gross motor skills, and other aspects compared to the control group. Skjoldborg et al. [26] also confirmed that FIVR has a positive effect on improving the life skills of children with ASD. Yuan et al. [7]. found that FIVR can effectively improve emotional expression and personal-social in children with ASD. The research results of the above-mentioned scholars are similar to this study, indicating that combined intervention has advantages in promoting the development of neuropsychological function and abilities in children with ASD [7, 26]. FIVR can stimulate the neuropsychological development of pediatric patients by creating specific training scenarios [7, 25, 26]. In terms of language training, VR environments can present vivid images, sounds, and interactive plots, attracting the attention of children and stimulating their desire to imitate and express themselves [7]. For the cultivation of adaptive behavior, FIVR can simulate various daily life scenarios, allowing children to repeatedly practice coping with different situations in a virtual environment, thereby improving their adaptive behavior in real life [27]. In terms of developing motor skills, some interactive games or training modules in FIVR can guide children to practice fine and gross movements. This novel training method is more effective in stimulating children’s interest and participation than traditional single psychological and behavioral interventions [27, 28]. In addition, Bioulac et al. [29] argue that skills and behaviors learned in VR environments can better generalize to real-life situations. Due to the fact that FIVR can simulate various real-life scenarios, the social, behavioral, and cognitive skills learned by children in virtual environments can be applied in similar real-life scenarios. This is of great significance for improving the overall function of children with ASD [27–29].

Several theoretical perspectives may help explain the observed benefits of fully immersive virtual reality (FIVR) for children with ASD. From a sensory integration perspective, FIVR provides a multi-sensory and controlled environment that enables individuals with autism to gradually adjust to external stimuli such as auditory input, facial expressions, and physical proximity, thereby promoting better sensory regulation [24]. Additionally, social motivation theory and mirror neuron system models suggest that interacting with lifelike virtual agents in a safe, simulated setting may enhance social engagement and imitation—key building blocks for communication skills. Ecological learning theory also supports the use of real-world context simulations (e.g., school, shopping, peer interaction) to increase generalization and adaptive functioning through repetitive practice in meaningful scenarios [25, 26]. Furthermore, FIVR facilitates embodied experiences by requiring users to navigate and interact using body movements and hand controllers, which may improve both gross and fine motor skills. These mechanisms collectively provide a neurodevelopmental rationale for integrating FIVR with behavioral interventions to support multi-domain improvements in ASD [27].

In addition, the results of this study showed that the satisfaction rate of family members of children in the FIVR group (95.2%) was higher than that of the control group (82.3%). This indicates that combining psychological and behavioral interventions with FIVR has significant advantages in improving the condition of pediatric patients, and the satisfaction of their families with this program is relatively high [28, 29]. The introduction of FIVR can bring more hope and confidence to family members. Family members can see that the child has made better progress under more innovative intervention modes, and thus be more satisfied with the intervention effect. Moreover, FIVR has significant effects in reducing the negative behaviors of children with autism and improving their abilities, which can reduce the pressure on family members when taking care of children with autism and thus improve satisfaction.

Given that this was a retrospective, non-randomized cohort study, causal inferences cannot be drawn from our findings. While associations between FIVR + PBI and various improvements were observed, these relationships should be interpreted with caution. Potential confounding factors, such as therapist variability or family involvement, cannot be fully ruled out. Therefore, our findings represent preliminary associations rather than definitive causal effects. These preliminary findings require confirmation through larger-scale, prospective randomized controlled trials to validate their generalizability and causal inference.

Building upon our findings, several future research directions are proposed to advance understanding and application of FIVR-based interventions in ASD: First, functional neuroimaging tools such as fMRI or fNIRS could be employed to examine brain activation patterns in response to FIVR, particularly in regions involved in social cognition, sensory processing, and executive function. These approaches would clarify the neural correlates underlying behavioral changes. Second, integrating physiological measures such as heart rate variability (HRV) and skin conductance during FIVR sessions may provide objective indicators of emotional regulation, stress response, and engagement, enhancing the precision of outcome assessments. Third, prospective randomized controlled trials (RCTs) with larger and more heterogeneous samples, along with long-term follow-up designs, are warranted to validate the causal effectiveness and durability of FIVR combined interventions. Lastly, future work may explore personalized VR content design by tailoring intervention modules based on individual cognitive profiles, behavioral phenotypes, or sensory sensitivities to improve treatment specificity and acceptability for diverse ASD subgroups. These avenues will support more robust, mechanism-informed, and individualized applications of immersive technologies in ASD rehabilitation.

This study has several limitations. Firstly, it was conducted as a single-center retrospective cohort study with a relatively small sample size, which may introduce selection bias and limit the generalizability of the findings. The retrospective, non-randomized design also imposes inherent methodological constraints, including limited ability to establish causal relationships and reduced internal validity. Although ANCOVA was applied to adjust for baseline covariates and diagnostic plots (Supplementary Figures S1–S6) confirmed that model assumptions were adequately met, residual confounding cannot be fully excluded, and the findings should still be interpreted with caution. Since data were extracted from existing clinical records rather than collected prospectively, there is a risk of incomplete documentation, inconsistency in intervention delivery, and lack of control over external variables. Secondly, no a priori power analysis was performed to determine the required sample size. As the study was retrospective in nature, the final cohort was composed of all eligible participants treated during the predefined observation period (January to July 2024). Although several outcome differences reached statistical significance, the absence of formal power estimation limits our ability to assess the risk of Type II error. Future prospective studies should include power calculations based on hypothesized effect sizes to ensure adequate statistical planning. Thirdly, several unmeasured factors may have influenced the observed outcomes. While all therapists were trained rehabilitation professionals operating under standardized protocols, individual differences in therapeutic style, clinical judgment, or communication may have contributed to variability in intervention delivery. Similarly, parental involvement, although encouraged as part of the program, was not quantitatively assessed. Variability in caregiver participation may have impacted children’s engagement and treatment responsiveness. In addition, we did not systematically collect data on co-interventions such as pharmacologic treatments, dietary supplementation, or concurrent educational programs, which could serve as confounding factors. Fourthly, the diagnostic procedures did not uniformly employ standardized diagnostic instruments such as the Autism Diagnostic Observation Schedule (ADOS) or the Social Communication Questionnaire (SCQ). While ASD diagnoses were made by qualified clinicians using DSM-5 criteria, the lack of formal diagnostic tools may have influenced classification consistency. Fifthly, the evaluation of intervention outcomes relied primarily on behavioral rating scales and observational performance. Objective neurobiological correlates, such as neuroimaging or electrophysiological indices, were not incorporated. This limits our ability to interpret the findings in relation to underlying neural mechanisms or plasticity effects. Sixthly, adverse reactions to immersive environments—such as cybersickness, fatigue, or disorientation—were not systematically recorded. The safety profile and tolerability of FIVR in children with ASD warrant systematic evaluation in future studies. Lastly, the study focused on short-term outcomes without long-term follow-up, making it difficult to determine the durability and sustained benefits of the intervention. Taken together, these limitations underscore the need for future prospective, multi-center randomized controlled trials that incorporate standardized diagnostic protocols, structured control of co-variables, objective neurophysiological endpoints, and long-term outcome tracking to validate and extend our preliminary findings.

Conclusion

This retrospective analysis suggests that the integration of fully immersive virtual reality (FIVR) with psychological and behavioral interventions may be associated with improvements in autism-related symptoms, neuropsychological functioning, and caregiver satisfaction in children with ASD. By incorporating ANCOVA adjusted for baseline covariates and reporting effect sizes with 95% confidence intervals, the robustness and clinical relevance of the observed group differences were further supported. Nevertheless, given the retrospective and non-randomized design, these findings remain preliminary. Future prospective randomized controlled trials are imperative to validate the effectiveness, durability, and underlying mechanisms of FIVR-based interventions across diverse clinical settings.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization: Ning Li, Lixin Sun and Boxia Li.Data curation: Maolin Tian and Yanfang Yang.Formal analysis: Zhenhuan Liu.Investigation: Ning Li, Lixin Sun and Boxia Li.Writing - review & editing: Ning Li, Lixin Sun and Boxia Li.

Funding information

None.

Data availability

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

Declarations

Ethical approval

All procedures performed in this study involving human participants were in accordance with the ethical standards of the institutional and/or national research committee, and the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. This study was approved by the Ethics Review Board of Tongren Hongxin Kangxin Traditional Chinese Medicine Hospital (No. 2024-1-1; November 13, 2024). The informed consent was waived by the Ethics Review Board of Tongren Hongxin Kangxin Traditional Chinese Medicine Hospital for the observational and retrospective nature.The authors declare no conflicts of interest that could be perceived as prejudicing the impartiality of the reported research.

Consent for publication

Not Applicable.

Competing interests

The authors declare no competing interests.