A Virtual Reality-Based Eye-Tracking System for Training Autistic Children

A Virtual Reality-Based Eye-Tracking System for Training Autistic Children | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article A Virtual Reality-Based Eye-Tracking System for Training Autistic Children Aiganym Soltiyeva, Wilk Oliveira, Madina Alimanova, Shyngys Adilkhan, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8852358/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Effective training of gaze-related behaviours in Autism Spectrum Disorder (ASD) requires systems capable of capturing and responding to visual attention with high temporal precision. However, existing virtual reality (VR) applications lack architectures for real-time, fine-grained gaze analysis, limiting their usefulness for adaptive training scenarios. To address this problem, this article presents the design, implementation, and evaluation of a VR-based eye-tracking system that enables fine-grained, real-time analysis of visual attention during interactions with virtual agents. The system incorporates a high-frequency gaze-event detection pipeline that uses RayCast-based sampling at 0.02-second intervals, a modular logging architecture for capturing temporal gaze metrics, and an adaptive task framework that dynamically responds to user attention. Implemented on the Oculus Quest Pro, the system records fixation distribution, gaze-to-agent transitions, and object-interaction patterns with millisecond-level precision. An experimental study with children with ASD demonstrates the system’s capability to accurately track gaze behaviour across multiple virtual characters and interactive elements. The results show the feasibility of using computational gaze modelling to support training-oriented VR environments for ASD. This work contributes a VR architecture for real-time gaze analysis and establishes a technical foundation for future intelligent, gaze-adaptive VR interventions. Virtual Reality Immersive technologies eye tracking Autism Spectrum Disorders assistive technologies Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 1 Introduction Virtual reality (VR) technologies are increasingly being explored as practical tools for supporting learning, assessment, and intervention for individuals with autism spectrum disorder (ASD) (G. G. Lorenzo et al., 2023; Silva et al., 2024; Koumpouros, 2025). Through immersive, controllable, and repeatable environments, VR facilitates the development of structured scenarios customized to diverse cognitive and sensory profiles (Karami et al., 2021; Soltiyeva et al., 2023; Damaševičius & Sidekerskienė, 2024; Sorrentino et al., 2025). Specifically, integration with eye-tracking technology is becoming a critical approach to improving virtual experiences and analysing user behaviour (Chen et al., 2025; Choi & Nam, 2025). Gaze tracking in VR environments facilitates interactive feedback with virtual content and provides valuable information about users’ attention and behavioural responses to particular objects or scenarios (Chen et al., 2025; Choi & Nam, 2025; Sorrentino et al., 2025). These characteristics make VR particularly well-suited to studying behaviour and delivering training in populations for whom real-world environments may be overwhelming or difficult to control (Creed et al., 2023). Despite significant advances in immersive technologies and computational interaction systems, developing robust, data-driven VR platforms that support users with diverse cognitive and sensory profiles remains a critical challenge in computer science (Creed et al., 2023; Dudley et al., 2023; Partarakis & Zabulis, 2024). Existing VR solutions for people with ASD often lack robust real-time processing pipelines, standardised gaze-based metrics, and adaptive mechanisms that can intelligently respond to dynamic user behaviour (Artiran et al., 2022b; Elkin et al., 2022b). In particular, while gaze-tracking sensors are increasingly integrated into modern head-mounted displays, many existing systems rely on offline gaze analysis, exhibit high latency, or do not incorporate gaze information into meaningful interaction logic (Elkin et al., 2022b; Yu et al., 2023). These limitations reduce VR environments’ ability to accurately model user engagement, capture attention patterns, and provide responsive, personalised learning experiences (Chen & Ren, 2025). To overcome these challenges, advances in computational design, real-time gaze analysis algorithms, and evidence-based evaluation protocols capable of detecting behavioural changes with high accuracy are needed. This paper proposes a VR-based gaze-tracking system to improve real-time communication and maintain eye contact in children with ASD. The proposed system integrates immersive VR environments with embedded gaze tracking, a high-frequency gaze-event detection mechanism, and a modular registration architecture to collect temporal gaze metrics. By continuously processing gaze data while interacting with virtual agents and objects, the system enables fine-grained analyses of visual attention and provides a technical framework for adapting attention-aware tasks in VR learning environments. The evaluation focuses on changes in gaze duration and engagement across multiple agents before and after VR-based training. Statistical analyses using paired-sample t-tests, Wilcoxon signed-rank tests, and effect-size estimation demonstrate statistically significant and practically meaningful improvements in visual attention following the intervention. This study aims to assess the effectiveness of a VR-based training system that integrates real-time eye-tracking and gaze-analysis algorithms. The study investigates whether this computational system can measurably alter user interaction patterns, with a focus on improvements in gaze behaviour and attention allocation among individuals with ASD. The objectives of this study are as follows: i) to develop an immersive VR-based training system equipped with a real-time eye-tracking feature; ii) to deploy and test the system with children diagnosed with ASD; iii) to define and compute novel system-level and gaze-based performance metrics; and iv) to evaluate the effectiveness of the system. The findings suggest that gaze-driven interaction can make VR environments more responsive to individual differences, creating a personalised learning experience that better meets each user's needs. Overall, this work offers a practical approach to integrating eye-tracking and immersive technologies and provides evidence that real-time gaze data reliably indicate engagement in VR-based training among individuals with ASD. 2 Background 2.1. Autism Spectrum Disorders Autism spectrum disorder (ASD) is a developmental disorder characterised by differences in social interaction and communication, variations in eye contact, and restricted interests or repetitive behaviours (Karami et al., 2021 ; Yang et al., 2024 ; Chen et al., 2025). Autism spectrum disorder is a condition with wide variation in the type and severity of signs, symptoms and levels of support needed by people with ASD. ASD is over four times more common among boys than girls, and it usually has co-occurring conditions, including epilepsy, depression, anxiety, and attention deficit hyperactivity disorder, as well as challenging behaviours such as sleep and self-harm (Wang et al., 2023 ). According to the Autism and Developmental Disabilities Monitoring statistics, approximately 1 in 31 children is diagnosed with ASD worldwide (Data and Statistics on Autism Spectrum Disorder, 2025). Differences in eye contact are often observed in the diagnostic profile of autism. Consequently, autistic individuals may demonstrate variations in social interaction and communication, reflecting the role of the eyes as a key source of social information that supports higher-order social-cognitive abilities, including theory of mind and perspective-taking (Stephenson et al., 2021 ; Stuart et al., 2022 ; Jeyarani & Senthilkumar, 2023 ). Maintaining stable eye contact can significantly enhance the quality of social experiences, increase the likelihood of appropriately responding to stimuli and cues, and contribute to the acquisition of adaptive social competencies (Alvari et al., 2021 ). 2.2. Virtual Reality The term “Virtual Reality” refers to the simulation and interaction with a realistic-looking world using computer graphics (Thakur et al., 2025; Schloss et al., 2025 ; Sorrentino et al., 2025 ). The ability to define and respond to user input, such as gestures and verbal commands, is a key feature called interactivity (Grasnick, 2021 ; Schloss et al., 2025 ). Interactivity and its captivating features contribute to a sense of immersion, allowing users to be and feel part of the device's environment (Grasnick, 2021 ; Schloss et al., 2025 ). Immersion is the ability of technology to create a sense of total presence and involvement in an artificial environment in which a digital world substitutes for reality (Jerald, 2015 ; Grasnick, 2021 ; Kaur & Josan, 2025 ; Thakur et al., 2025). Human–computer interaction (HCI) and information and communication technologies (ICTs) are undergoing a transition, driven by innovative design paradigms that reshape how we interact with digital information. Immersive technologies such as augmented reality (AR), virtual reality (VR), and mixed reality (MR) are increasingly regarded as the building blocks of contemporary digital experiences. Human-computer interaction (HCI) in VR is underway through specialised interfaces that allow users to input commands to the computer and receive feedback from the simulation. Nowadays, to address humans' sensory channels, VR applications and devices vary in functionality and purpose (Burdea & Coiffet, 2003 ). The history of virtual reality (VR) dates back to the 18th century, beginning with the invention of the first stereoscopic 3D TV by Sir Charles Wheatstone. David Brewster’s handheld stereoscope, demonstrated in 1851, paved the way for the development of VR technology, along with Morton Heilmeier’s head-mounted display (HMD) and Sensorama. Ivan Sutherland's “The Ultimate Display,” presented in 1960, was a fundamental blueprint for VR. The early XXI century is known as a “VR winter”, when VR’s potential was explored in depth by government, academy and military research laboratories around the globe (Jerald, 2015 ; LaValle, 2023 ). VR is a powerful tool for the entertainment industry, especially for gaming and immersive filmmaking (Liu et al., 2022 ; H. Li et al., 2022 ). Beyond entertainment, VR has been successfully utilised across industries, including oil and gas, education, medicine, design, therapy, the military, and flight training (Rojas-Sánchez et al., 2022 ; C, 2024 ). Utilising VR technologies is cost-effective, as it avoids mistakes before manufacturing, saves time by speeding up iterative processes, provides safe and ecologically valid environments for training, and visualises large datasets that would be incomprehensible with traditional systems (Burdea & Coiffet, 2003 ; Joshi et al., 2020 ; Soltiyeva et al., 2023 ). Extended Reality, or XR, is a collective term for immersive technologies that combine the physical and digital worlds, encompassing Virtual Reality (VR), Augmented Reality (AR), and Mixed Reality (MR). Augmented Reality (AR) is the concept of digitally overlaying virtual objects onto real-world objects, allowing users to interact with them simultaneously. AR graphics are visible on smartphones, tablets, and other devices, providing a new form of interactive experience for users. Mixed Reality (MR), a combination of AR and VR, blends the physical and digital worlds (Guo et al., 2021 ; Morimoto et al., 2022 ; Schwaiger et al., 2024 ). Research indicates that transforming educational methodologies through VR, AR, and XR technologies provides immersive digital experiences, interactive environments, simulations, and enhanced engagement (Al-Ansi et al., 2023 ). However, VR technologies remain in their early stages of development, and there is a gap in their swift implementation and customisation within educational institutions. 2.3. VR in the education of children with ASD Implementing VR technologies in education has several advantages, providing more exploratory, engaging, and playful learning experiences. One of the most important capabilities of VR is the provision of a personalised learning experience and feedback tailored to each student's needs, enabling students to explore VR environments, collaborate with peers worldwide, and make the experience more active and interactive (Marougkas et al., 2023 ). In contrast to traditional learning settings, VR technologies offer safe, cost-effective, and time-saving experiences (Rojas-Sánchez et al., 2022 ). Furthermore, integration of VR technologies in educational settings has shown a positive impact on student learning, academic performance, motivation and engagement. The primary reason for these advantages is that they enable VR students to discover, explore, and interact with objects in ways that are not possible in the physical world (Al-Ansi et al., 2023 ). The increasing number of individuals with ASD, along with their growing interest in computerised programs, has prompted numerous studies to develop technologies such as robotics, VR/AR via headsets and goggles, interactive video modelling, and mobile or touchpad devices (Karami et al., 2021 ). VR technologies provide diverse capabilities to meet the needs of both typical users and individuals with special needs (Mesa-Gresa et al., 2018 ). Primarily, VR-based training systems can provide a safe, ecologically valid and personalised environment for the individual needs of the ASD population. Because autistic individuals often exhibit differences in peer communication, traditional school settings may present social and environmental demands that can be stressful without appropriate support. In such cases, VR-based learning systems serve as an additional tool for training and educating children with ASD, providing real-life scenarios in a less anxious and more manageable environment (Yuan & Ip, 2018 ; Zhang et al., 2022 ). VR technologies, when combined with gamified approaches, can increase motivation, attention, and focus among participants with ASD (Shahab et al., 2021 ). Intervention programmes using VR techniques allow repetitive practice by repeatedly presenting tasks and stimuli, an important aspect of interventions for autistic individuals (Didehbani et al., 2016 ). Current research covers a wide range of training interventions, including social interaction and communication skills, emotional skills, daily living skills, and cognitive functions (Karami et al., 2021 ). 2.4. Related Work Eye-tracking technology has long been available for desktop displays; however, in recent years, several commercial solutions have emerged that enable eye tracking on consumer VR head-mounted displays (HMDs). Consequently, research on eye tracking in HMDs has accelerated and expanded significantly (Adhanom et al., 2023 ). Learning and training are important applications of VR, and combining it with eye tracking holds great promise for assessing skills and learning outcomes and for improving training (Smutny, 2022 ; Adhanom et al., 2023 ; Altın et al., 2025 ). Devices such as the Meta Quest Pro, HTC Vive Pro Eye, Varjo VR-3 and Pico Neo 3 Pro Eye allow developers to capture gaze direction, fixation points, and pupil indices with high temporal and spatial accuracy (González & Bozkir, 2024 ). Recent developments in computer science have increasingly focused on combining eye-tracking technology with VR to better understand and support social and attentional behaviours in autistic individuals (Chen et al., 2025). One of the earliest examples of this integration was presented by Lahiri et al. ( 2011 ), who designed a gaze-sensitive VR system that provided adaptive feedback based on real-time gaze data, demonstrating that eye-tracking could be used effectively to train social interaction skills. Elkin, Zhang, and Reneker ( 2022 ) developed the VR SAFE project, which features a virtual classroom designed to provide a social context for training activities. Using the HTC Vive Pro Eye VR headset, the researchers tracked participants’ eye gaze and measured how often they looked into the avatar's eyes. Sanku et al. ( 2023 ) developed a VR classroom focused on TEDx-style talks, in which each participant shares personal life incidents. The HMD and wristband collected participants’ real-time eye-gaze behaviour and heart rate data. Razzak et al. ( 2024 ) utilised a VR classroom environment with the HTC Vive Pro Eye headset to manipulate and measure participants’ attention and engagement. Pino et al. ( 2021 ) conducted an eye-tracking–based comparative study analysing visual attention patterns of children with autism spectrum condition (ASC) when observing real versus virtual avatar faces expressing basic emotions. Using a Tobii eye-tracking system, the authors quantified gaze behaviour by measuring total fixation duration and the number of fixations, and conducted survival analyses of exploration patterns within manually defined areas of interest (eyes and mouth). Table 1 Comparison of the related studies # Study Eye-tracking Number of participants Summary and relevance 1 Design of a Gaze-Sensitive Virtual Social Interactive System for Children With Autism (Lahiri et al., 2011 ) The Virtual Interactive system with Gaze-sensitive Adaptive Response Technology (VIGART) using VR-based social situation as a platform for delivering individualized feedback based on one’s dynamic gaze patterns. 6 The findings indicate that participants' physiological eye responses during VR-based social interactions presented in VIGART may reveal whether they can recognise emotions, consistent with observations from non-VR-based tasks. Thus, there is reason to believe that VIGART could be used as an intervention tool, potentially supporting children with autism spectrum disorder in improving their social skills. 2 Gaze Fixation and Visual Searching Behaviors during an Immersive Virtual Reality Social Skills Training Experience for Children and Youth with Autism Spectrum Disorder: A Pilot Study (Elkin, Zhang & Reneker, 2022 ). The eye tracker on the HTC Vive Pro Eye VR tracked how often the participant looked into the eyes of the avatar, and for how long the participant looked in the virtual environment. 11 The study's findings on gaze fixation and visual search behaviour during simulated training revealed patterns in individuals with mild-to-moderate ASD. The use of this objective measure of gaze fixation behaviour during social skills training may be helpful for similar intervention applications and provide a more accurate measure of individual tendencies in social interaction. 3 Enhancing attention in autism spectrum disorder: comparative analysis of virtual reality-based training programs using physiological data(Sanku et al., 2023 ). The HTC Vive Pro tracks eye information, including pupil size and gaze direction. 25 This study provides a comprehensive evaluation of the Virtual Reality Physiological Data Analysis system, investigating the relationship between physiological data and attention in individuals. Regarding eye-tracking data analysis, the findings provide evidence of improved attention in individuals with ASD. 4 Using virtual reality to enhance attention for autistic spectrum disorder with eye tracking (Razzak et al., 2024 ). The HTC VIVE Eye Pro headset, eye-tracking, eye openness, pupil size and gaze duration. 50 The findings indicate the effectiveness of VR as a tool for enhancing attention management in individuals with ASD, highlighting the importance of personalised therapeutic interventions. 5 Comparing virtual vs real faces expressing emotions in children with autism: An eye-tracking study (Pino et al., 2021 ) Tobii T120 Eye Tracker equipment consisting of a GL-2760-LED backlit monitor with a resolution of 1920×1080 pixels, which tracks both eyes to an accuracy of 0.5 degrees at a sampling rate of 60 Hz. 29 The results confirm that children with ASD have higher capacities to process and recognize emotions when these are presented by avatar faces. Together, these studies demonstrate the promise of VR and eye-tracking technologies for both understanding and supporting individuals with ASD. However, most existing work remains confined to observational or diagnostic applications or to short-term pilot studies. Few systems include real-time gaze analysis with adaptive feedback or establish clear quantitative measures of behavioural improvement. Addressing these limitations, the present study introduces a computationally grounded VR-based training system that leverages real-time eye-tracking and gaze-analysis algorithms to enhance social attention and evaluate measurable training outcomes. 3 Concept and Design of the VR System 3.1. Instructional Design Principles The system, called My Lovely Granny’s Farm , simulates a farm environment containing household objects, animals, and two interactive virtual characters—a farmer and his son. Users can freely navigate the environment, explore and investigate virtual elements, respond to questions, and complete task-oriented activities. The system integrates real-time eye-tracking to continuously capture users’ gaze behaviour, including fixation duration, attention shifts between virtual agents, and gaze–object interactions during task execution. These gaze data are used both for behavioural analysis and to support adaptive interaction within the virtual environment. Given that prior research has demonstrated the positive effects of animal-based and nature-inspired environments on the mental health and emotional regulation of individuals with impairments (Berry et al., 2016 ; Rodriguez et al., 2019 ; O’Haire & Rodriguez, 2021 ), the farm setting was intentionally selected for this study. During training, participants interacted with the virtual farmer and the young boy, and their visual attention to socially relevant cues—such as faces, gestures, and task-related objects—was objectively monitored via eye tracking. This design enables the assessment of social attention patterns alongside active communication practice. Figure 2 illustrates the overall system architecture and data flow. Virtual farm with all objects, with virtual characters created in Autodesk Maya 2022. The ready scene, including all necessary 3D models and animated characters, was exported to Unity in .fbx format. The Oculus Quest Pro VR headset is used for immersive interaction within virtual environments. The virtual characters, the farmer and son, and the VR farm are demonstrated in Fig. 3 . 3.2. Implementation In this VR-based learning system, characters interact with children using pre-recorded lines. Arm, leg, and torso movements and gestures are included, giving the characters more natural, realistic behaviour and facilitating better user interaction. For more precise interaction with the VR headset, the official Meta libraries were used for camera control and movement rather than the Unity universal libraries. A separate animation accompanies each utterance. Character animations are created in Autodesk Maya and built in Unity using AnimatorController. The AnimatorController uses the State Machine pattern. Animations in this system are considered States, where States are switched through Transitions under specific Conditions. The Conditions vary depending on the Parameters. Figure 4 shows the Component Audio Source. The option to use the controllers independently has been removed to lower the threshold for children entering VR. It is also possible to move the controllers in space while remaining in place, but this is not recommended for children due to the risk of dizziness. The operator controls the entire session scenario manually via controllers outside the VR zone. Below is an audio source script: v o i d S t a r t ( ) { F a r m e r a u d i o = GetComponent ( ) ; } v o i d h e l l o P l a y ( ) { F a r m e r a u d i o. PlayOneShot ( h e l l o ) ; } v o i d h a n d P l a y ( ) { F a r m e r a u d i o. PlayOneShot ( showhand ) ; } v o i d g o o d J o b P l a y ( ) { F a r m e r a u d i o. PlayOneShot ( g o o d j o b ) ; } v o i d i a m H e r e P l a y ( ) { F a r m e r a u d i o. PlayOneShot ( i a m h e r e ) ; } For more precise interaction with the VR headset, the official Meta libraries were used for camera control and movement rather than the Unity universal libraries. The game has two characters, and their animations and replicas are switched through the Scenario Controller script – the operator Button.One button switches the index of the required animation for the Farmer character; when the index reaches the end of the list, it wraps around to the beginning. The Button.Two buttons start the animation of the Farmer's character with the current pointer (index). All of this is performed by the right controller. The same logic applies to the boy character (Aidos). Only Button.Three is used to switch the pointer and Button.Four on the left controller to start the animation. Controller mapping is demonstrated in Fig. 5 below. 3.3. Eye Tracking Integration and Logging The Meta Quest Pro has advanced inward-facing infrared (IR) cameras and near-infrared LED illuminators for real-time eye tracking. These cameras capture high-resolution grayscale images of the eyes, allowing the system to detect and monitor pupil position, eye blinks, and gaze direction with high temporal precision (approximately 90–120 Hz). The underlying eye-tracking process combines traditional model-based gaze estimation with data-driven neural network models. The typical stages include: Pupil detection using image processing techniques such as ellipse fitting or Hough Circle Transform, or more robustly, via Convolutional Neural Networks (CNNs) trained on large eye image datasets. 3D gaze vector estimation using a geometric model of the human eye, calculating the vector from the corneal center through the pupil to determine gaze direction. Machine learning-based gaze regression directly maps eye image features to screen coordinates or world-space vectors, thereby increasing robustness to user variability (particularly beneficial for children with Autism Spectrum Disorder). The Meta Quest Pro combines these approaches to achieve enhanced accuracy (≈ 1°–2°) and low latency (< 50 ms), enabling real-time, gaze-based interaction within immersive virtual environments. Eye-tracking functionality was integrated into the VR application using Meta's Presence Platform SDK in Unity. The following process was implemented: SDK Setup: The Oculus Integration package (version 51 or later) was imported into the Unity project. The EyeTracking feature was enabled in Project Settings > XR Plugin Management. Scripting Access: Eye-tracking data was accessed via Unity C# scripts: v a r e y e T r a c k i n g D a t a = OVRPlugin. G e t E y e T r a c k i n g S t a t e ( ) ; V e c t o r 3 g a z e D i r e c t i o n = e y e T r a c k i n g D a t a. E y e G a z e D i r e c t i o n ; V e c t o r 3 g a z e O r i g i n = e y e T r a c k i n g D a t a. E y e G a z e O r i g i n ; These vectors were used to perform ray casts to detect which object the user was looking at. Object Identification: Each interactive object in the scene (e.g., Farmer, Cow, Tree) was tagged and registered with a unique identifier (name). The object's name was retrieved if the gaze ray intersected a collider. Fixation Duration: A timer was implemented to detect fixations (gaze durations > 500 ms), which indicate intentional focus rather than transient glances. To capture gaze behavior during training sessions, a custom logging system was developed to store fixation data in JSON format: { ”name ” : ” Farmer ” , ” t i m e ” : ” 0, 5 8 ” , ” d a t e ” : ” 2 2. 3 2. 1 0 1 9. 0 2. 2 5 ” } Each entry logs the following: “name”: The name of the object fixated upon; “time”: Duration of the gaze fixation (in seconds); “date”: Timestamp of the fixation (hh.mm.ss dd.mm.yy). name = “Farmer” indicates that the user is fixated on an NPC character labelled “Farmer.” time = “0,58” indicates that the user looked at the object for 0.58 seconds (exceeding the 500 ms threshold). date = “22.32.10 19.02.25” provides the time and date of the fixation event (22:32:10 on February 19, 2025). The script collects these entries during runtime and writes them to a structured log file at session end or predefined intervals for later analysis of attention patterns, learning behaviour, and interaction frequency. Figure 6 demonstrates the design and workflow of the VR-based eye-tracking training system. 4 Methods 4.1. Study Design This study employed a within-subject experimental design to explore how a VR system with integrated eye-tracking can support training for individuals with ASD. The goal was to understand how gaze-based interaction can be used to capture patterns of visual attention and enable adaptive system responses during structured VR tasks. Each participant engaged in a predefined set of virtual activities while their gaze behaviour was continuously monitored and recorded. This design enabled consistent observation of user interaction, allowing reliable comparisons while reducing the influence of external factors. The eye-tracking component is not in itself a learning intervention. Rather, learning is delivered through structured VR tasks and social interaction scenarios, and eye tracking is used to enable objective observation and analysis. Specifically, the integrated eye tracking system captures gaze behaviour, including fixation duration, gaze transitions, and attention allocation to virtual agents and task-relevant objects. This data is used to identify attention patterns and user interaction behaviour, which serve as measurable indicators of engagement and social attention during training. Thus, gaze tracking in this system plays a dual role: (i) it supports the quantification of user behaviour by providing accurate gaze metrics, and (ii) it provides adaptive system logic by allowing the VR environment to respond to detected attention states when necessary. It is important to note that the behavioural improvements observed in this study are attributable to the VR training design, with gaze tracking serving as a measurement and feedback layer that enhances the system's ability to assess, analyse, and personalise learning. 4.2 Participants The experimental procedures lasted from February to April 2025, and 10 children aged 3–6 (8 boys and 2 girls), 10 parents, and 9 specialists from the children's correctional centre participated. The training sessions were conducted at the children’s correctional centre Dinamica City , which provides services for children with special educational needs and various developmental diagnoses, including autism, intellectual disability with autistic features, Down syndrome, and others. The centre accommodates children aged 12 months to 17 years, with a maximum capacity of approximately 60. Dinamica City was established in 2021 by Zhunussova Lyazat Kaimbekovna, who serves as both the director and a therapist specialising in ASD. A memorandum of cooperation was signed with the centre, and written consent was obtained from parents permitting their children to participate in the training sessions. Participants received detailed instructions on using the equipment and were introduced to the system's fundamental principles. The primary aim of this stage was to improve the children's communication skills, including eye contact and emotional expression. Table 3 shows participants' characteristics, i.e., age and diagnosis. Table 3 ‒ Participants' characteristics ID Age Gender Diagnosis ASD 1 6 male Autism, group 4, according to O.S. Nikolskaya's classification ASD 2 5 female Autism, group 2, according to O.S. Nikolskaya's classification ASD 3 8 male Cerebral palsy, moderate intellectual disability with autistic features ASD 4 6 male Autism, group 2, according to O.S. Nikolskaya's classification ASD 5 7 male Atypical autism with intellectual disability ASD 6 4 male Autism, group 4, according to O.S. Nikolskaya's classification ASD 7 7 female Autism, group 2, according to O.S. Nikolskaya's classification ASD 8 4 male Delayed psycho-verbal development, level 2 general speech underdevelopment. ASD 9 7 male Cerebral palsy, mild intellectual disability. General speech underdevelopment at level 1 ASD 10 7 male Autism, group 2, according to O.S. Nikolskaya's classification All participants received instructions for using the Meta Quest Pro VR headset and a session on adapting to the virtual environment. They then underwent observational and interaction-based assessments in a structured VR environment prior to exposure to adaptive learning content. To ensure consistency and comfort during the session, participants were allowed a brief familiarisation period within the virtual environment before the assessments. This period enabled them to explore fundamental interactions and navigate the space, thereby reducing potential anxiety or disorientation. During this time, facilitators monitored their responses and provided minimal guidance to encourage independent engagement while ensuring safety and understanding of the virtual setting. 4.3. Materials The study employed a virtual reality system with an integrated eye-tracking module. Participants interacted with the system via a head-mounted display that captured real-time gaze information, including fixation points and gaze direction. A custom VR environment was developed to present interactive tasks that elicit and sustain visual attention. Within this environment, predefined areas of interest were used to support detailed gaze-based analysis. The VR application was run on a desktop computer, which also recorded interaction data. In addition, dedicated software components were implemented to log gaze coordinates, timestamps, and interaction events for later analysis. 4.4 Measures The assessment measured time spent interacting (in seconds) with seven in-game virtual agents: Farmer, Boy, Horse, Chicken, Dog, Sheep, and Cow . Additionally, two key behavioural indicators were recorded: the number of initiated conversations (discrete count) and response time (in seconds). These measures were selected as proxy indicators of engagement (attention duration to virtual agents), social initiation, and information-processing time, which are significant problems in autism and key targets for intervention. A structured parent-report questionnaire was developed to assess changes in the social functioning of children with ASD before and after the VR training intervention. The questionnaire (APPENDIX A) consisted of 9 social skill items, each rated on a 10-point Likert scale, with 1 indicating minimal proficiency and 10 indicating consistent, independent demonstration of the skill. These metrics included maintaining eye contact, initiating peer conversation, joining group play without prompts, turn-taking, personal hygiene, expressing empathy, recognising facial expressions, responding to greetings, and introducing oneself. Parents provided ratings for each skill before and after the intervention, enabling a paired-comparison analysis. Additionally, a “Brief Description” field was included for each item to collect qualitative observational data, thereby providing contextual depth. A final open-ended section invited parents to reflect on the effectiveness of the VR system, capturing user perceptions and anecdotal evidence. The combination of quantitative behavioural data from the VR environment and qualitative insights from parent reports aimed to provide a multidimensional understanding of the intervention's impact. By triangulating immersive interaction metrics with caregiver observations, the study sought to evaluate not only immediate engagement and responsiveness within the virtual context but also the potential for skill generalisation to real-world social settings. This mixed-methods approach allowed for a more comprehensive evaluation of behavioural change, capturing both measurable improvements and nuanced parental perspectives on developmental progress following the VR training. Table 4 shows eye-tracking and behavioural metrics used to evaluate the VR-based training system. Table 4 Overview of target stimuli and associated gaze- and behaviour-based measurements Target stimulus Type of data Function Farmer Gaze duration (seconds) Measures visual attention directed toward a primary human virtual agent during interaction Boy Gaze duration (seconds) Assesses attention allocation toward a peer-like virtual character Horse Gaze duration (seconds) Evaluates visual attention to non-human animated agents within the environment Chicken Gaze duration (seconds) Measures engagement with secondary animated objects and agents Dog Gaze duration (seconds) Assesses attention to socially relevant animal agents Sheep Gaze duration (seconds) Captures the distribution of gaze toward the background animated elements Cow Gaze duration (seconds) Evaluates visual attention to non-human animated agents within the environment Initiated Conversations Number of conversations Quantifies the frequency of social interaction initiated by the user Response time Time duration (seconds) Measures the latency of user responses to prompts or social cues Social skills questionnaire (SSQ ) Score Provides an external behavioural measure to contextualise in-VR gaze and interaction outcomes 4.5. Procedure Before the training sessions began, all participants received safety instructions to ensure a secure and comfortable experience with the VR system. These included guidance on proper use of the VR headset, awareness of the physical boundaries of the training area to prevent collisions or falls, and instructions to notify supervisors immediately if discomfort or dizziness occurs. A trained staff member was present throughout the session to monitor participants and assist as needed. Figure 7 demonstrates the floor plan of the training area. A parent-reported questionnaire was administered to assess social skills and eye contact in autistic individuals prior to VR-based eye-tracking training. Before the experimental tasks began, the VR head-mounted display was fitted and adjusted for comfort. An eye-tracking calibration procedure was then conducted to align the system with each participant’s visual characteristics and ensure accurate gaze measurement. Calibration involved directing the participant’s attention to a series of visual targets within the virtual environment and was repeated if necessary to achieve acceptable tracking accuracy. Following calibration, participants completed a set of predefined VR tasks designed to elicit visual attention and interaction with virtual elements. Figure 8 provides an overview of the study procedure. When users wear the Oculus Pro headset, they appear in a virtual farm populated with domestic animals, trees, vegetables, and two virtual characters. Users can walk, explore, and interact with virtual objects and the environment. Participants must complete tasks designed to improve their social skills and communication. Depending on the maintenance of eye contact with each virtual agent (e.g., a farmer, a boy, or animals), the system fixes the duration of eye gaze for each participant. Table 5 presents a description of the tasks. Table 5 ‒ Description of the tasks Task Questions Feedback Greeting What is your name? My name is Aidos. How old are you? Great, I am seven years old Social interaction How are you? I want to show you my grandmother's farm. How are you feeling? Do you like a horse? Do you want to come closer to the horse? During task execution, the system continuously recorded gaze behaviour and interaction events. Participants were encouraged to proceed at their own pace, and short breaks were provided when needed to minimise fatigue or discomfort. At the end of the session, the VR equipment was removed, and participants were given time to rest. All recorded data were securely stored for subsequent preprocessing and analysis. After finishing the experimental part, parents filled out post-questionaire to fix changes in social skills and eye-contact. 4.6. Data analysis A quantitative approach was employed to obtain a comprehensive understanding of participants’ responses to the VR-based training system. To evaluate the effectiveness of the proposed VR-based training system, statistical analyses were selected based on the data structure, sample characteristics, and research objectives. Since gaze-related metrics were collected from the same participants before and after the intervention, paired-sample statistical tests were required to assess within-subject changes attributable to the training. A paired-samples t-test was employed to compare pre- and post-training measurements when the data met the assumptions of approximate normality. The paired t -test is appropriate in this context because it is specifically designed to detect mean differences between two related conditions and provides high statistical power for small-to-moderate sample sizes commonly observed in experimental studies involving children with ASD. Given the potential for non-normal distributions and increased variability in behavioural data, particularly in neurodiverse populations, the Wilcoxon signed-rank test was additionally applied as a non-parametric alternative. This test does not assume normality and is therefore well suited to validating results when distributional assumptions may be violated. The use of both parametric and non-parametric tests strengthens the robustness and reliability of the findings. To complement statistical significance testing, effect size measures (Cohen’s d ) were calculated to quantify the magnitude of observed changes. While p -values indicate whether an effect is statistically detectable, effect sizes provide essential information about the practical and clinical relevance of the intervention, which is especially important in applied VR and autism research. Together, this multi-layered analytical approach ensures that the evaluation captures both statistical reliability and practical significance, offering a comprehensive assessment of the system’s impact on gaze behaviour and user engagement. Engagement was measured using rating scales assessing time-on-task, while the VR system automatically captured eye-tracking data via detailed interaction logs. Each log entry recorded the name of the object or virtual agent the participant focused on (e.g., “Farmer”), the duration of gaze in seconds (e.g., 0.58 s), and a timestamp indicating the exact time of the interaction. These data allowed for fine-grained analysis of visual attention, including fixation durations, gaze transitions, and interactions with virtual elements. Longer gaze durations were interpreted as higher interest or engagement with the content. For the qualitative analysis, changes in social interaction and emotional responses were assessed using video recordings and field notes collected before and after VR training sessions. Additionally, parents completed the Social Skills Questionnaire (SSQ) to provide standardised assessments of participants’ social skills pre- and post-intervention. Observations from the initial sessions informed iterative improvements to the VR environment, including the introduction of new objects and animals, as well as additional tasks and dialogue prompts designed to enhance engagement and communication skills. This method design enabled triangulation of data sources, combining objective gaze metrics, standardised assessments, and observational insights to provide a robust evaluation of the system’s effectiveness in improving attention, engagement, and social behaviour in children with ASD. 5 Results This study evaluated changes in visual attention toward virtual agents following interaction with the proposed VR-based eye-tracking training system. Gaze behaviour was quantified using fixation duration and gaze engagement metrics derived from the real-time RayCast-based eye-tracking pipeline. Paired-sample t -tests were conducted to compare pre-intervention and post-intervention gaze metrics for each virtual agent. Given the small sample size ( N = 11) and potential non-normality of the gaze data, Wilcoxon signed-rank tests were also performed to assess the robustness of the results. Effect sizes were calculated using Cohen’s d to determine the magnitude of the observed changes. Across all virtual agents, mean gaze duration increased from the pre-training phase to the post-training phase. At baseline, participants demonstrated limited visual engagement with both social and task-related agents. Following the VR-based training sessions, gaze duration and fixation frequency increased consistently, indicating enhanced visual attention toward the agents embedded in the virtual environment. Although individual variability was observed, the overall trend across participants reflected improved attentional engagement after exposure to the VR system. Paired-sample t -test results revealed statistically significant increases in gaze duration for all evaluated virtual agents following the intervention ( p < 0.05 for all comparisons). The most substantial statistical effects were observed for agents with guiding or interactive roles (e.g., Farmer and Chicken), indicating that agent function and behavioural salience influenced attention allocation. To confirm these findings, Wilcoxon signed-rank tests were performed, yielding significant pre–post differences for all agents ( p = 0.001), supporting the consistency and reliability of the parametric results. As illustrated in Fig. 10 , mean gaze duration increased across all virtual agents following the VR-based training. The most pronounced increases were observed for socially salient agents such as the Farmer and Boy. Cohen’s d values indicated large effect sizes across all agents, with values ranging from large to very large according to conventional thresholds ( d ≥ 0.8), with the most potent effects observed for the Farmer and Chicken agents. The largest effect sizes were associated with socially salient or instruction-oriented agents, while the smallest observed effect still exceeded the threshold for a significant impact. These findings suggest that the VR-based training system produced not only statistically significant but also practically meaningful improvements in visual attention behaviour. Figure 11 demonstrates the Cohen’s d effect sizes for changes in gaze duration toward virtual agents following VR-based eye-tracking training. Figure 12 shows the changes in mean gaze duration before and after training for all virtual agents. A consistent upward trend is observed across all agents, indicating increased visual attention following VR-based gaze-tracking training. While baseline gaze duration was similarly low for all agents, post-training values increased markedly, especially for socially relevant and learning-oriented agents such as Farmer, Boy, and Chicken. These trajectories indicate that the observed improvements are systematic rather than agent-specific anomalies. The uniform direction of change across agents corroborates the statistical evidence and suggests that the intervention resulted in consistent within-subject improvements in visual attention. The results demonstrate that the proposed VR-based eye-tracking system consistently improved visual attention across all evaluated virtual agents. Analysis of pre- and post-training gaze data showed a systematic increase in mean gaze duration, as evidenced by statistically significant paired-sample t -tests and confirmed by Wilcoxon signed-rank tests. Effect size analysis indicated large practical effects across agents, demonstrating that the observed changes were robust and meaningful. Together, these findings provide clear evidence that the VR-based training intervention effectively enhanced gaze engagement through immersive, eye-tracking–driven interaction. 6 Discussion The objective of this study was to investigate whether an immersive VR system with integrated eye-tracking could effectively support visual attention training in individuals with ASD. The results provide strong empirical evidence that the proposed system significantly enhanced gaze engagement across a range of virtual agents. Increases in gaze duration and fixation engagement following the intervention indicate that immersive VR environments can facilitate meaningful improvements in attentional processes in children with ASD. At baseline, participants exhibited limited spontaneous gaze toward virtual agents, which is consistent with previous research documenting atypical attention allocation in ASD. Following VR-based training, participants demonstrated increased and more sustained visual attention, suggesting that repeated interaction within a structured, predictable, and low-anxiety virtual environment promoted attentional regulation. The most significant effects were associated with agents that played explicit social or guiding roles, highlighting the importance of agent design, behavioural clarity, and interaction logic in capturing and maintaining attention within VR-based interventions. The findings of this study align with and extend prior work on VR-based interventions for ASD. For example, Lorenzo et al. ( 2016 ) demonstrated that immersive VR environments support the development of social understanding and daily living skills by providing structured and repeatable interaction scenarios. Similarly, the significant improvements observed in social engagement metrics and SSQ scores in the present study suggest that immersive VR facilitates safe and effective practice of social behaviours. Li et al. ( 2025 ) reported improvements in attention and reduced response times in task-based VR interventions for children with ASD, findings consistent with the decreased response times observed in this study, suggesting more efficient cognitive processing during post-intervention sessions. Furthermore, the present results complement the systematic review by Mesa-Gresa et al. ( 2018 ), which highlighted the potential of immersive, gamified VR applications to improve sensory integration, emotional regulation, and communication skills in individuals with ASD. Improvements in interactions with virtual characters, such as the Farmer, Boy, Chicken, and Dog, suggest increased familiarity, comfort, and engagement in socially meaningful role-play contexts. The substantial improvement in SSQ scores further indicates that immersive VR may positively influence perceived social competence and readiness, consistent with findings reported by Ke and Moon ( 2018 ) in avatar-mediated learning environments. Compared with prior studies employing non-immersive or partially immersive systems, the fully immersive VR environment used in this study may account for the more consistent and pronounced improvements across multiple behavioural and cognitive measures. This observation supports the hypothesis that immersion plays a critical role in enhancing engagement and facilitating deeper learning, as suggested by Newbutt et al. ( 2019 ). In contrast to studies reporting moderate statistical effects, the proposed system consistently demonstrated strong effects across all key metrics, underscoring the intervention's robustness. In contrast to earlier studies using 2D video or static image stimuli (e.g., Chita-Tegmark et al., 2016; Falck-Ytter et al., 2013 ), the present study utilised a fully immersive VR environment featuring seven distinct agents. Unlike prior findings suggesting a preference for non-social stimuli, children in this study exhibited increased gaze duration toward human-like agents, such as the Farmer and Boy, particularly in post-test sessions (Chita-Tegmark et al., 2016; Falck-Ytter et al., 2013 ). This aligns with Chevallier et al. ( 2015 ), who emphasised the role of realism in promoting social gaze, and Wang et al. (2018), who reported enhanced attention to interactive virtual agents. These results suggest that immersive, context-rich environments may mitigate common attentional biases in ASD and promote more naturalistic social orientation patterns (Chita-Tegmark, 2015 ). Beyond confirming prior findings, this study contributes to the field by integrating high-frequency, real-time eye-tracking into the core interaction logic of an immersive VR system. By combining computational gaze modelling with adaptive task design, the proposed architecture enables attention-responsive training rather than passive assessment. This positions eye-tracking not only as an evaluation tool but as an active mechanism for guiding interaction and learning in VR-based interventions for ASD. 6.1 Limitations of the Study While this study provides encouraging insights, several limitations should be acknowledged. First, the number of participants was relatively small, which may limit the extent to which the findings can be generalised. Autism spectrum disorder encompasses a wide range of cognitive and behavioural characteristics, and a larger, more diverse sample would likely reveal additional patterns and variations in gaze behaviour that were not captured in this study. Another important consideration relates to the technical constraints of eye-tracking technology. Although the system was designed to operate in real time, its accuracy may be affected by factors such as headset calibration, head motion, lighting conditions, and individual differences in eye physiology. These factors may introduce noise into gaze data and degrade the precision of attention measurements, even when the system operates as intended. The experimental setup itself also introduces certain limitations. The virtual environment was intentionally structured to support controlled observation and consistent interaction; however, this level of control may not fully reflect how users behave in more complex or naturalistic settings. As a result, the extent to which the observed behaviours generalise to real-world contexts remains an open question. In addition, the study primarily relied on gaze-based indicators to infer attention and engagement. While gaze provides valuable insights into visual focus, it represents only one aspect of cognitive and emotional processing. Incorporating additional modalities—such as facial expression analysis, physiological signals, or performance-based measures—could offer a more comprehensive understanding of user experience. Finally, the study focused on short-term interactions and did not examine long-term use or learning effects. Future research should explore how user behaviour and engagement evolve over extended periods and whether repeated exposure leads to sustained improvements. Addressing these limitations in future work will help strengthen the robustness and applicability of VR-based eye-tracking systems for individuals with ASD. 6.2. Practical implications Using immersive VR technologies to improve an individual's performance expands understanding of personalised education and therapy models. This is consistent with cognitive-behavioural theories that support personalised interventions, reinforcing the idea that different individuals need unique ways of learning. IVRS for ASD children has given rise to new theories of Human-Computer Interaction, investigating the interaction of children with neurodevelopmental disorders with digital environments. The unique ways in which autistic individuals process sensory information have informed more nuanced HCI models, thereby stimulating the development of accessible and engaging interfaces for children with support needs. Utilising IVRS to simulate different environmental stimuli provides practical insights into sensory processing theories. The findings suggest that controlled sensory exposure in a virtual environment can help children with ASD manage and improve their responses to sensory input, thereby contributing to existing sensory integration and adaptation theories. IVRS contributes to the theory of therapeutic environment design, particularly by elucidating the features that make virtual environments useful for behavioural interventions. The customizable, structured, and predictable nature of VR environments aligns with cognitive-behavioural therapy and deepens theoretical understanding of how environments influence treatment effectiveness. VR-based learning systems have provided autistic children with a practical, repetitive approach to practising social interactions in a secure, controlled environment. Simulating different social scenarios allows children to practice recognising social cues, understanding appropriate responses, and improving conversational skills. Collaboration among computer scientists, VR developers, educators, and healthcare professionals has led to an interdisciplinary approach to problem-solving in the treatment of ASD. It fosters innovation in software development for neurodiverse populations and leads to the creation of VR applications designed to achieve therapeutic outcomes. These contributions highlight how IVRS advances theoretical models in computer science, education, and therapy and offers scalable, practical solutions that can significantly improve the quality of life of autistic individuals. 6.3. Conclusion The application of IVRS for autistic individuals demonstrates substantial potential for enhancing social, cognitive, and behavioural skills, particularly when combined with integrated eye-tracking technologies. The findings of this study show that the proposed IVRS provides an engaging and interactive learning environment that captures and sustains the visual attention of children with ASD more effectively than traditional instructional approaches. By incorporating real-time eye-tracking, the system enables objective measurement of gaze behaviour, fixation patterns, and attention allocation during interaction with virtual agents, offering fine-grained insight into user engagement and learning processes. The multisensory nature of VR, together with gaze-based interaction and feedback, supports sustained attention and focus, thereby enhancing the retention and reinforcement of learned skills. Eye-tracking further enables adaptive system behaviour by allowing the virtual environment to respond dynamically to individual attention patterns, facilitating personalised learning experiences tailored to each child’s abilities and sensory preferences. In addition, the controlled and safe nature of VR environments, combined with continuous gaze monitoring, allows children to practise social interactions, interpret social cues, and regulate sensory responses without exposure to real-world risks or overstimulation. Through repeated exposure to gaze-guided social scenarios, children can practise attention to socially relevant stimuli, such as virtual characters and interactive objects, thereby supporting the gradual transfer of these skills to real-life communication contexts. From an empirical perspective, the effectiveness of the proposed eye-tracking–enabled IVRS was supported by systematic quantitative analysis of pre- and post-intervention gaze metrics. Paired-sample t -tests revealed statistically significant improvements in visual attention across multiple virtual agents, while complementary robustness checks confirmed the consistency of these findings. Effect size analysis indicated large practical effects, demonstrating that the observed changes were not only statistically reliable but also meaningful for user engagement and attention regulation. Collectively, the integration of eye-tracking analytics with immersive VR interaction provides strong evidence that data-driven, gaze-adaptive virtual environments can serve as an effective and scalable intervention framework for supporting social learning in autistic individuals. 6.4. Future Research Directions Future research will focus on conducting quantitative studies with larger sample sizes to investigate the emotional impact of VR experiences on children with ASD. By collecting and analysing quantitative data, researchers will gain more statistically significant insights into how VR systems affect emotions such as anxiety, joy, or frustration. These studies will use emotion-detection tools, such as facial expression analysis and physiological measures (e.g., heart rate and skin conductance), to objectively assess children's emotional reactions during and after VR sessions. Another pivotal area of future research will be the development and implementation of new features in the VR system and the improvement of personalisation. Personalised interactions for each child's unique needs, preferences, and educational needs will be considered. Future systems may include advanced algorithms that automatically adapt scenarios and tasks, or provide real-time feedback based on a child's behaviour, thereby offering a more engaging and personalised experience. This could include customised avatars, customizable environments, and adaptive learning pathways. One intended enhancement is the implementation of voice recognition technology to generate more natural and interactive dialogue in VR environments. This will allow children to interact verbally with virtual characters, helping them practice language and social skills in real time. Integrating advanced speech-recognition tools will make interactions more responsive and personalised, potentially improving the communication skills of autistic children. Lastly, future research should investigate how children perceive and respond to changes in the system, including individualised graphic elements tailored to each participant. These elements may include personalised visual stimuli, colours, or themes aligned with a child's interests or sensory preferences. Investigating how children perceive these changes will provide insight into the effectiveness of personalisation and its impact on user engagement, emotional responses, and learning outcomes. Declarations Acknowledgements We sincerely thank all the professionals and researchers who contributed to this study. Funding This work has been supported by the Academy of Finland Flagship Programme [Grant No. 337653 - Forest-Human-Machine Interplay (UNITE)]. Authors’ contributions Aiganym Soltiyeva: Conceptualisation, Methodology, Formal analysis, Validation, Writing - review & editing. Wilk Oliveira: Supervision, Writing - review & editing. Alimanova Madina: Supervision, Writing - review & editing. Shyngys Adilkhan: Writing - review & editing Marat Urmanov: Writing - review & editing Conflicts of interest There are no conflicts of interest to declare. Ethics declarations The study was approved by the Research Ethics Committee of SDU University. Written informed consent was obtained from the parents or legal guardians of all participants prior to participation. Consent for publication Written informed consent was obtained from the parents or legal guardians for the publication of any potentially identifiable images or data included in this article. Availability of data and material (data transparency) The original dataset available as supplementary material Code availability Not applicable References Adhanom, I. B., MacNeilage, P., & Folmer, E. (2023). Eye Tracking in Virtual Reality: a Broad Review of Applications and Challenges. Virtual Reality , 27 (2), 1481–1505. https://doi.org/10.1007/s10055-022-00738-z Al-Ansi, A. M., Jaboob, M., Garad, A., & Al-Ansi, A. (2023). Analyzing augmented reality (AR) and virtual reality (VR) recent development in education. Social Sciences & Humanities Open, 8(1), 100532. https://doi.org/10.1016/j.ssaho.2023.100532 Altın, Y., Boşnak, Ö., & Turhan, C. (2025). Examining Virtual Reality Interventions for Social Skills in Children with Autism Spectrum Disorder: A Systematic Review. Journal of Autism and Developmental Disorders . https://doi.org/10.1007/s10803-025-06741-y Alvari, G., Coviello, L., & Furlanello, C. (2021). EYE-C: Eye-Contact Robust Detection and Analysis during Unconstrained Child-Therapist Interactions in the Clinical Setting of Autism Spectrum Disorders. Brain Sciences , 11 (12), 1555. https://doi.org/10.3390/brainsci11121555 Artiran, S., Ravisankar, R., Luo, S., Chukoskie, L., & Cosman, P. (2022b). Measuring Social Modulation of Gaze in Autism Spectrum Condition With Virtual Reality Interviews. IEEE Transactions on Neural Systems and Rehabilitation Engineering , 30 , 2373–2384. https://doi.org/10.1109/tnsre.2022.3198933 Burdea, G., & Coiffet, P. (2003). Virtual Reality Technology. PRESENCE Virtual and Augmented Reality , 12 (6), 663–664. https://doi.org/10.1162/105474603322955950 Berry, A., Borgi, M., Francia, N., Alleva, E., & Cirulli, F. (2016). Effectiveness of animal-assisted interventions: A systematic review of randomized controlled trials. International Journal of Environmental Research and Public Health, 13(6), Article 618. https://pubmed.ncbi.nlm.nih.gov/24731910/ C, A. B. (2024). ENHANCING MILITARY TRAINING THROUGH VR APPLICATIONS. International Scientific Journal of Engineering and Management , 03 (05), 1–9. https://doi.org/10.55041/isjem01739 Chevallier C., Parish-Morris J., Tonge N. et al. Susceptibility to distraction in autism spectrum disorder: Probing the integrity of attentional control using eye tracking // Frontiers in Human Neuroscience. – 2015. – Vol. 9. – P. 124-1-124-7. Chen, H., & Ren, Y. (2025). Auxiliary diagnostic method for children with autism spectrum disorder based on virtual reality and eye-tracking technology. Scientific Reports , 15 (1), 40552. https://doi.org/10.1038/s41598-025-24243-w Chita-Tegmark, M. (2015). Social attention in ASD: A review and meta-analysis of eye-tracking studies. Research in Developmental Disabilities , 48 , 79–93. https://doi.org/10.1016/j.ridd.2015.10.011 Choi, H., & Nam, S. (2025). Effect of eye-tracking-based depth perception visualization interface on virtual reality user experience. Virtual Reality , 30 (1). https://doi.org/10.1007/s10055-025-01260-8 Creed, C., Al-Kalbani, M., Theil, A., Sarcar, S., & Williams, I. (2023). Inclusive AR/VR: accessibility barriers for immersive technologies. Universal Access in the Information Society , 23 (1), 59–73. https://doi.org/10.1007/s10209-023-00969-0 Data and statistics on autism spectrum disorder . (2025, May 27). Autism Spectrum Disorder (ASD). https://www.cdc.gov/autism/data-research/index.html Didehbani, N., Allen, T., Kandalaft, M., Krawczyk, D., & Chapman, S. (2016). Virtual Reality Social Cognition Training for children with high functioning autism. Computers in Human Behavior , 62 , 703–711. https://doi.org/10.1016/j.chb.2016.04.033 Dudley, J., Yin, L., Garaj, V., & Kristensson, P. O. (2023). Inclusive Immersion: a review of efforts to improve accessibility in virtual reality, augmented reality and the metaverse. Virtual Reality , 27 (4), 2989–3020. https://doi.org/10.1007/s10055-023-00850-8 Elkin, T. D., Zhang, Y., & Reneker, J. C. (2022). Gaze Fixation and Visual Searching Behaviors during an Immersive Virtual Reality Social Skills Training Experience for Children and Youth with Autism Spectrum Disorder: A Pilot Study. Brain Sciences , 12 (11), 1568. https://doi.org/10.3390/brainsci12111568 Falck-Ytter T., Bölte S., Gredebäck G. Eye tracking in early autism research // J of Neurodevelopmental Disorders. – 2013. – Vol. 5, Issue 1. – P. 28-1-28-13. González, N. M., & Bozkir, E. (2024). Eye-Tracking Devices for Virtual and Augmented Reality Metaverse Environments and Their Compatibility with the European Union General Data Protection Regulation. Digital Society , 3 (2). https://doi.org/10.1007/s44206-024-00128-9 Grasnick, A. (2021). Basics of virtual reality . https://doi.org/10.1007/978-3-662-64201-6 Guo, X., Guo, Y., & Liu, Y. (2021). The Development of Extended Reality in Education: Inspiration from the Research Literature. Sustainability , 13 (24), 13776. https://doi.org/10.3390/su132413776 Jerald, J. (2015). The VR Book: Human-Centered Design for Virtual Reality . https://doi.org/10.1145/2792790 Jeyarani, R. A., & Senthilkumar, R. (2023). Eye Tracking Biomarkers for Autism Spectrum Disorder Detection using Machine Learning and Deep Learning Techniques: Review. Research in Autism Spectrum Disorders , 108 , 102228. https://doi.org/10.1016/j.rasd.2023.102228 Joshi, S., Hamilton, M., Warren, R., Faucett, D., Tian, W., Wang, Y., & Ma, J. (2020). Implementing Virtual Reality technology for safety training in the precast/ prestressed concrete industry. Applied Ergonomics , 90 , 103286. https://doi.org/10.1016/j.apergo.2020.103286 Karami, B., Koushki, R., Arabgol, F., Rahmani, M., & Vahabie, A. (2021). Effectiveness of Virtual/Augmented Reality–Based therapeutic Interventions on Individuals with Autism Spectrum Disorder: A Comprehensive Meta-Analysis. Frontiers in Psychiatry , 12 . https://doi.org/10.3389/fpsyt.2021.665326 Kaur, B., & Josan, G. S. (2025). Bringing Faces to Life: A survey on realistic facial expressions in 3D virtual characters. International Journal of Human-Computer Interaction , 1–42. https://doi.org/10.1080/10447318.2025.2561766 Ke, F., & Moon, J. (2018). Virtual collaborative gaming as social skills training for high‐functioning autistic children. British Journal of Educational Technology , 49 (4), 728–741. https://doi.org/10.1111/bjet.12626 Koumpouros, Y. (2025). Digital Horizons: Enhancing Autism Support with Augmented Reality. Journal of Autism and Developmental Disorders . https://doi.org/10.1007/s10803-024-06709-4 Lahiri, U., Warren, Z., & Sarkar, N. (2011). Design of a Gaze-Sensitive Virtual Social interactive system for children with autism. IEEE Transactions on Neural Systems and Rehabilitation Engineering , 19 (4), 443–452. https://doi.org/10.1109/tnsre.2011.2153874 LaValle, S. M. (2023). Virtual reality . Cambridge university press. Li, H., Lo, C., Smith, A., & Yu, Z. (2022). The development of virtual production in film industry in the past decade. In Lecture notes in computer science (pp. 221–239). https://doi.org/10.1007/978-3-031-06047-2_16 Li, N., Tian, M., Yang, Y., Liu, Z., Sun, L., & Li, B. (2025). The effect of fully immersive virtual reality technology combined with psychological and behavioral intervention on autism spectrum disorder. BMC Psychology , 13 (1), 1120. https://doi.org/10.1186/s40359-025-03460-y Liu, Z., Jin, Y., Ma, M., & Li, J. (2022). A comparison of Immersive and Non-Immersive VR for the education of filmmaking. International Journal of Human-Computer Interaction , 39 (12), 2478–2491. https://doi.org/10.1080/10447318.2022.2078462 Lorenzo, G., Lledó, A., Pomares, J., & Roig, R. (2016). Design and application of an immersive virtual reality system to enhance emotional skills for children with autism spectrum disorders. Computers & Education , 98 , 192–205. https://doi.org/10.1016/j.compedu.2016.03.018 Lorenzo, G. G., Newbutt, N. N., & Lorenzo-Lledó, A. A. (2023). Designing virtual reality tools for students with Autism Spectrum Disorder: A systematic review. Education and Information Technologies , 28 (8), 9557–9605. https://doi.org/10.1007/s10639-022-11545-z Marougkas, A., Troussas, C., Krouska, A., & Sgouropoulou, C. (2023). Virtual Reality in Education: A review of learning theories, approaches and methodologies for the last decade. Electronics , 12 (13), 2832. https://doi.org/10.3390/electronics12132832 Mesa-Gresa, P., Gil-Gómez, H., Lozano-Quilis, J., & Gil-Gómez, J. (2018). Effectiveness of Virtual Reality for Children and Adolescents with Autism Spectrum Disorder: An Evidence-Based Systematic Review. Sensors , 18 (8), 2486. https://doi.org/10.3390/s18082486 Morimoto, T., Kobayashi, T., Hirata, H., Otani, K., Sugimoto, M., Tsukamoto, M., Yoshihara, T., Ueno, M., & Mawatari, M. (2022). XR (Extended Reality: Virtual Reality, Augmented Reality, Mixed Reality) Technology in Spine Medicine: status quo and Quo vadis. Journal of Clinical Medicine , 11 (2), 470. https://doi.org/10.3390/jcm11020470 Meta for Developers. (2025, November 10). Get started with basic controller input in Unity. Meta. https://developers.meta.com/horizon/documentation/unity/unity-tutorial-basic-controller-input/ Newbutt, N., Bradley, R., & Conley, I. (2019). Using Virtual Reality Head-Mounted Displays in Schools with Autistic Children: Views, Experiences, and Future Directions. Cyberpsychology Behavior and Social Networking , 23 (1), 23–33. https://doi.org/10.1089/cyber.2019.0206 O’Haire, M. E., & Rodriguez, K. E. (2021). Animal-assisted activities improve social behaviors in psychiatrically hospitalized youth with autism. Journal of Applied Developmental Psychology, 72, Article 101272. https://pubmed.ncbi.nlm.nih.gov/30818971/ Partarakis, N., & Zabulis, X. (2024). A review of immersive Technologies, knowledge representation, and AI for Human-Centered Digital Experiences. Electronics , 13 (2), 269. https://doi.org/10.3390/electronics13020269 Pino, M. C., Vagnetti, R., Valenti, M., & Mazza, M. (2021). Comparing virtual vs real faces expressing emotions in children with autism: An eye-tracking study. Education and Information Technologies , 26 (5), 5717–5732. https://doi.org/10.1007/s10639-021-10552-w Razzak, R., Li, Y., He, J., Jung, S., Mei, C., & Huang, Y. (2024). Using virtual reality to enhance attention for autistic spectrum disorder with eye tracking. High-Confidence Computing , 5 (1), 100234. https://doi.org/10.1016/j.hcc.2024.100234 Rodriguez, K. E., Greer, J., Yatcilla, J. K., Beck, A. M., & O’Haire, M. E. (2019). The effects of assistance dogs on psychosocial health and wellbeing: A systematic literature review. PLOS ONE, 14(8), e0256071. https://doi.org/10.1371/journal.pone.0256071 Rojas-Sánchez, M. A., Palos-Sánchez, P. R., & Folgado-Fernández, J. A. (2022). Systematic literature review and bibliometric analysis on virtual reality and education. Education and Information Technologies , 28 (1), 155–192. https://doi.org/10.1007/s10639-022-11167-5 Sanku, B. S., Li, Y., Jung, S., Mei, C., & He, J. (2023). Enhancing attention in autism spectrum disorder: comparative analysis of virtual reality-based training programs using physiological data. Frontiers in Computer Science , 5 . https://doi.org/10.3389/fcomp.2023.1250652 Shahab, M., Taheri, A., Mokhtari, M., Shariati, A., Heidari, R., Meghdari, A., & Alemi, M. (2021). Utilizing social virtual reality robot (V2R) for music education to children with high-functioning autism. Education and Information Technologies , 27 (1), 819–843. https://doi.org/10.1007/s10639-020-10392-0 Schloss, I., Dillon, A., & Bailey, J. (2025). Why use virtual reality as a treatment for children and youth’s mental health and wellbeing: a review. Digital Society , 4 (2). https://doi.org/10.1007/s44206-025-00193-8 Schwaiger, M., Krajnčan, M., Vukovič, M., Jenko, M., & Doz, D. (2024). Educators’ opinions about VR/AR/XR: An exploratory study. Education and Information Technologies . https://doi.org/10.1007/s10639-024-12808-7 Silva, R. M., Martins, P., & Rocha, T. (2024). Virtual reality educational scenarios for students with ASD: Instruments validation and design of STEM programmatic contents. Research in Autism Spectrum Disorders , 119 , 102521. https://doi.org/10.1016/j.rasd.2024.102521 Smutny, P. (2022). Learning with virtual reality: a market analysis of educational and training applications. Interactive Learning Environments , 31 (10), 6133–6146. https://doi.org/10.1080/10494820.2022.2028856 Soltiyeva, A., Oliveira, W., Madina, A., Adilkhan, S., Urmanov, M., & Hamari, J. (2023). My Lovely Granny’s Farm: An immersive virtual reality training system for children with autism spectrum disorder. Education and Information Technologies , 28 (12), 16887–16907. https://doi.org/10.1007/s10639-023-11862-x Sorrentino, G., Edwards, M. G., Baldini, N., Mustile, M., Lejeune, T., & Everard, G. (2025). REAsmash-ET: a methodological framework for combined cognitive and motor assessment through eye-tracking and kinematic metrics in immersive VR search-and-reach task. Journal of NeuroEngineering and Rehabilitation , 23 (1), 25. https://doi.org/10.1186/s12984-025-01844-0 Stephenson, L. J., Edwards, S. G., & Bayliss, A. P. (2021). From gaze perception to social cognition: the Shared-Attention System. Perspectives on Psychological Science , 16 (3), 553–576. https://doi.org/10.1177/1745691620953773 Stuart, N., Whitehouse, A., Palermo, R., Bothe, E., & Badcock, N. (2022). Eye gaze in Autism Spectrum Disorder: A review of Neural Evidence for the eye Avoidance hypothesis. Journal of Autism and Developmental Disorders , 53 (5), 1884–1905. https://doi.org/10.1007/s10803-022-05443-z Wang, L., Wang, B., Wu, C., Wang, J., & Sun, M. (2023). Autism spectrum Disorder: neurodevelopmental risk factors, biological mechanism, and precision therapy. International Journal of Molecular Sciences , 24 (3), 1819. https://doi.org/10.3390/ijms24031819 Yang, X., Wu, J., Ma, Y., Yu, J., Cao, H., Zeng, A., Fu, R., Tang, Y., & Ren, Z. (2024). Effectiveness of virtual reality technology intervention in improving social skills of children and adolescents with autism:A systematic review (Preprint). Journal of Medical Internet Research . https://doi.org/10.2196/60845 Yuan, S. N. V., & Ip, H. H. S. (2018). Using virtual reality to train emotional and social skills in children with autism spectrum disorder. London Journal of Primary Care , 10 (4), 110–112. https://doi.org/10.1080/17571472.2018.1483000 Yu, C., Wang, S., Zhang, D., Zhang, Y., Cen, C., You, Z., Zou, X., Deng, H., & Li, M. (2023, October 20). HSVRS: a virtual reality system of the Hide-and-Seek game to enhance gaze fixation ability for autistic children . arXiv.org. https://arxiv.org/abs/2310.13482 Zhang, M., Ding, H., Naumceska, M., & Zhang, Y. (2022). Virtual Reality Technology as an Educational and Intervention Tool for Children with Autism Spectrum Disorder: Current Perspectives and Future Directions. Behavioral Sciences , 12 (5), 138. https://doi.org/10.3390/bs12050138 Additional Declarations No competing interests reported. Supplementary Files APPENDIXA.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8852358","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":627428782,"identity":"11c336b2-71d9-4173-9c4e-cff2d8b99f45","order_by":0,"name":"Aiganym Soltiyeva","email":"data:image/png;base64,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","orcid":"","institution":"SDU University","correspondingAuthor":true,"prefix":"","firstName":"Aiganym","middleName":"","lastName":"Soltiyeva","suffix":""},{"id":627428785,"identity":"3485f388-870a-42a2-b8ae-e4982eab07ad","order_by":1,"name":"Wilk Oliveira","email":"","orcid":"","institution":"Tampere University","correspondingAuthor":false,"prefix":"","firstName":"Wilk","middleName":"","lastName":"Oliveira","suffix":""},{"id":627428789,"identity":"9fedfa05-c99d-41c7-84aa-82d5864cdc4b","order_by":2,"name":"Madina Alimanova","email":"","orcid":"","institution":"SDU University","correspondingAuthor":false,"prefix":"","firstName":"Madina","middleName":"","lastName":"Alimanova","suffix":""},{"id":627428793,"identity":"6624f567-4e6f-40ea-99bd-89ea5d63631b","order_by":3,"name":"Shyngys Adilkhan","email":"","orcid":"","institution":"SDU University","correspondingAuthor":false,"prefix":"","firstName":"Shyngys","middleName":"","lastName":"Adilkhan","suffix":""},{"id":627428796,"identity":"22820891-658f-4c16-a419-34d5c9e538be","order_by":4,"name":"Marat Urmanov","email":"","orcid":"","institution":"SDU University","correspondingAuthor":false,"prefix":"","firstName":"Marat","middleName":"","lastName":"Urmanov","suffix":""}],"badges":[],"createdAt":"2026-02-11 13:38:31","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8852358/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8852358/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":107834532,"identity":"25994ad8-6374-4a06-8bd5-6f03140537ee","added_by":"auto","created_at":"2026-04-26 15:46:01","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":5713,"visible":true,"origin":"","legend":"\u003cp\u003eLegend not included with this version\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8852358/v1/a70045a3d3ff1e8ee47384ec.png"},{"id":107870378,"identity":"dca69034-67c5-4554-9518-c6a24ae308b9","added_by":"auto","created_at":"2026-04-27 07:39:32","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1157621,"visible":true,"origin":"","legend":"\u003cp\u003eThe system diagram\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8852358/v1/aa0c3d045ae6d13e402ed6b7.png"},{"id":107834535,"identity":"7e43db9b-8a70-423c-ab2d-9c5602442d8f","added_by":"auto","created_at":"2026-04-26 15:46:01","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":464851,"visible":true,"origin":"","legend":"\u003cp\u003eVirtual characters\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8852358/v1/7339b6c77d63fdba7cac7b0f.jpeg"},{"id":108006623,"identity":"356b6ab6-073d-4c66-91bd-01d67e63fa3d","added_by":"auto","created_at":"2026-04-28 12:56:11","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":289202,"visible":true,"origin":"","legend":"\u003cp\u003eComponent Audio Source\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8852358/v1/46301b383a1bd424d2a6696e.jpeg"},{"id":107834537,"identity":"75aad68e-5e0b-4fa6-8287-f487fbb590d3","added_by":"auto","created_at":"2026-04-26 15:46:01","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":166695,"visible":true,"origin":"","legend":"\u003cp\u003eController mapping (Meta for Developers, 2025)\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8852358/v1/a79a56508f90725d19ab7ed1.png"},{"id":107870397,"identity":"1d8f730f-8537-435d-b553-4f71dec9334c","added_by":"auto","created_at":"2026-04-27 07:39:35","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":248693,"visible":true,"origin":"","legend":"\u003cp\u003eDesign and Workflow of the VR Eye-Tracking Training System\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8852358/v1/2cce8d667b2e29d7901e7d99.png"},{"id":107869565,"identity":"ae3b3003-dec3-4d22-b8ba-71209e19ffc4","added_by":"auto","created_at":"2026-04-27 07:37:26","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":20814,"visible":true,"origin":"","legend":"\u003cp\u003eFloor plan of the training area\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8852358/v1/d930b3678c0145c1fc4709df.png"},{"id":107834539,"identity":"be38dd92-77f8-4a48-ba5a-b173e96a1242","added_by":"auto","created_at":"2026-04-26 15:46:01","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":41941,"visible":true,"origin":"","legend":"\u003cp\u003eProcedure overview\u003c/p\u003e","description":"","filename":"floatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-8852358/v1/b12131c4fe42c325382a7f9d.png"},{"id":107834541,"identity":"08828b66-36d9-4ec7-81ba-863f4dd86074","added_by":"auto","created_at":"2026-04-26 15:46:01","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":43694,"visible":true,"origin":"","legend":"\u003cp\u003eMean gaze duration toward VR agents before and after training\u003c/p\u003e","description":"","filename":"floatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-8852358/v1/b608c5494c5fddb740da99db.png"},{"id":107870060,"identity":"1adf4063-6205-4130-ab12-8a312b67ff09","added_by":"auto","created_at":"2026-04-27 07:38:44","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":5713,"visible":true,"origin":"","legend":"\u003cp\u003eLegend not included with this version\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-8852358/v1/a836c6d8da0488ace106e4ce.png"},{"id":107834543,"identity":"fa58a433-ce08-475a-85a4-4bd75c9a9b28","added_by":"auto","created_at":"2026-04-26 15:46:01","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":60850,"visible":true,"origin":"","legend":"\u003cp\u003eCohen’s \u003cem\u003ed\u003c/em\u003e effect sizes for changes in gaze duration\u003c/p\u003e","description":"","filename":"floatimage9.png","url":"https://assets-eu.researchsquare.com/files/rs-8852358/v1/584a37f7772b4728f5c1d2b6.png"},{"id":107870490,"identity":"d6b19238-fe2a-4605-845a-dbc3d54a704e","added_by":"auto","created_at":"2026-04-27 07:39:46","extension":"png","order_by":12,"title":"Figure 12","display":"","copyAsset":false,"role":"figure","size":119428,"visible":true,"origin":"","legend":"\u003cp\u003ePre- to Post-Training changes in gaze duration across virtual agents\u003c/p\u003e","description":"","filename":"floatimage10.png","url":"https://assets-eu.researchsquare.com/files/rs-8852358/v1/8b10ef60f0ebc8be32a11447.png"},{"id":108803691,"identity":"faa89c56-2b01-4c8c-a0bb-9fad008f30c4","added_by":"auto","created_at":"2026-05-08 15:03:52","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2880873,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8852358/v1/fa683980-bf49-4e52-bcd9-2547abedba7e.pdf"},{"id":107870206,"identity":"16fcecc9-600a-4d65-bf37-71341086eb29","added_by":"auto","created_at":"2026-04-27 07:39:08","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":17108,"visible":true,"origin":"","legend":"","description":"","filename":"APPENDIXA.docx","url":"https://assets-eu.researchsquare.com/files/rs-8852358/v1/e773654b6a03f5f3a55873c4.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"A Virtual Reality-Based Eye-Tracking System for Training Autistic Children","fulltext":[{"header":"1 Introduction","content":"\u003cp\u003eVirtual reality (VR) technologies are increasingly being explored as practical tools for supporting learning, assessment, and intervention for individuals with autism spectrum disorder (ASD) (G. G. Lorenzo et al., 2023; Silva et al., 2024; Koumpouros, 2025). Through immersive, controllable, and repeatable environments, VR facilitates the development of structured scenarios customized to diverse cognitive and sensory profiles (Karami et al., 2021;\u0026nbsp;Soltiyeva et al., 2023; Damaševičius \u0026amp; Sidekerskienė, 2024; Sorrentino et al., 2025). Specifically, integration with eye-tracking technology is becoming a critical approach to improving virtual experiences and analysing user behaviour (Chen et al., 2025; Choi \u0026amp; Nam, 2025). Gaze tracking in VR environments facilitates interactive feedback with virtual content and provides valuable information about users’ attention and behavioural responses to particular objects or scenarios (Chen et al., 2025; Choi \u0026amp; Nam, 2025; Sorrentino et al., 2025). These characteristics make VR particularly well-suited to studying behaviour and delivering training in populations for whom real-world environments may be overwhelming or difficult to control (Creed et al., 2023).\u003c/p\u003e\n\u003cp\u003eDespite significant advances in immersive technologies and computational interaction systems, developing robust, data-driven VR platforms that support users with diverse cognitive and sensory profiles remains a critical challenge in computer science (Creed et al., 2023; Dudley et al., 2023; Partarakis \u0026amp; Zabulis, 2024). Existing VR solutions for people with ASD often lack robust real-time processing pipelines, standardised gaze-based metrics, and adaptive mechanisms that can intelligently respond to dynamic user behaviour (Artiran et al., 2022b; Elkin et al., 2022b). In particular, while gaze-tracking sensors are increasingly integrated into modern head-mounted displays, many existing systems rely on offline gaze analysis, exhibit high latency, or do not incorporate gaze information into meaningful interaction logic (Elkin et al., 2022b; Yu et al., 2023). These limitations reduce VR environments’ ability to accurately model user engagement, capture attention patterns, and provide responsive, personalised learning experiences (Chen \u0026amp; Ren, 2025). To overcome these challenges, advances in computational design, real-time gaze analysis algorithms, and evidence-based evaluation protocols capable of detecting behavioural changes with high accuracy are needed.\u003c/p\u003e\n\u003cp\u003eThis paper proposes a VR-based gaze-tracking system to improve real-time communication and maintain eye contact in children with ASD. The proposed system integrates immersive VR environments with embedded gaze tracking, a high-frequency gaze-event detection mechanism, and a modular registration architecture to collect temporal gaze metrics. By continuously processing gaze data while interacting with virtual agents and objects, the system enables fine-grained analyses of visual attention and provides a technical framework for adapting attention-aware tasks in VR learning environments.\u0026nbsp;The evaluation focuses on changes in gaze duration and engagement across multiple agents before and after VR-based training. Statistical analyses using paired-sample t-tests, Wilcoxon signed-rank tests, and effect-size estimation demonstrate statistically significant and practically meaningful improvements in visual attention following the intervention.\u003c/p\u003e\n\u003cp\u003eThis study aims to assess the effectiveness of a VR-based training system that integrates real-time eye-tracking and gaze-analysis algorithms. The study investigates whether this computational system can measurably alter user interaction patterns, with a focus on improvements in gaze behaviour and attention allocation among individuals with ASD. The objectives of this study are as follows: i) to develop an immersive VR-based training system equipped with a real-time eye-tracking feature; ii) to deploy and test the system with children diagnosed with ASD; iii) to define and compute novel system-level and gaze-based performance metrics; and iv) to evaluate the effectiveness of the system.\u003c/p\u003e\n\u003cp\u003eThe findings suggest that gaze-driven interaction can make VR environments more responsive to individual differences, creating a personalised learning experience that better meets each user's needs. Overall, this work offers a practical approach to integrating eye-tracking and immersive technologies and provides evidence that real-time gaze data reliably indicate engagement in VR-based training among individuals with ASD.\u003c/p\u003e"},{"header":"2 Background","content":"\u003cdiv id=\"Sec2\" class=\"Section2\"\u003e \u003ch2\u003e2.1. Autism Spectrum Disorders\u003c/h2\u003e \u003cp\u003eAutism spectrum disorder (ASD) is a developmental disorder characterised by differences in social interaction and communication, variations in eye contact, and restricted interests or repetitive behaviours (Karami et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Yang et al., \u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Chen et al., 2025). Autism spectrum disorder is a condition with wide variation in the type and severity of signs, symptoms and levels of support needed by people with ASD. ASD is over four times more common among boys than girls, and it usually has co-occurring conditions, including epilepsy, depression, anxiety, and attention deficit hyperactivity disorder, as well as challenging behaviours such as sleep and self-harm (Wang et al., \u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). According to the Autism and Developmental Disabilities Monitoring statistics, approximately 1 in 31 children is diagnosed with ASD worldwide (Data and Statistics on Autism Spectrum Disorder, 2025). Differences in eye contact are often observed in the diagnostic profile of autism. Consequently, autistic individuals may demonstrate variations in social interaction and communication, reflecting the role of the eyes as a key source of social information that supports higher-order social-cognitive abilities, including theory of mind and perspective-taking (Stephenson et al., \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Stuart et al., \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Jeyarani \u0026amp; Senthilkumar, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Maintaining stable eye contact can significantly enhance the quality of social experiences, increase the likelihood of appropriately responding to stimuli and cues, and contribute to the acquisition of adaptive social competencies (Alvari et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.2. Virtual Reality\u003c/h2\u003e \u003cp\u003eThe term \u0026ldquo;Virtual Reality\u0026rdquo; refers to the simulation and interaction with a realistic-looking world using computer graphics (Thakur et al., 2025; Schloss et al., \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Sorrentino et al., \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). The ability to define and respond to user input, such as gestures and verbal commands, is a key feature called interactivity (Grasnick, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Schloss et al., \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Interactivity and its captivating features contribute to a sense of immersion, allowing users to be and feel part of the device's environment (Grasnick, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Schloss et al., \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Immersion is the ability of technology to create a sense of total presence and involvement in an artificial environment in which a digital world substitutes for reality (Jerald, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Grasnick, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Kaur \u0026amp; Josan, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Thakur et al., 2025).\u003c/p\u003e \u003cp\u003eHuman\u0026ndash;computer interaction (HCI) and information and communication technologies (ICTs) are undergoing a transition, driven by innovative design paradigms that reshape how we interact with digital information. Immersive technologies such as augmented reality (AR), virtual reality (VR), and mixed reality (MR) are increasingly regarded as the building blocks of contemporary digital experiences. Human-computer interaction (HCI) in VR is underway through specialised interfaces that allow users to input commands to the computer and receive feedback from the simulation. Nowadays, to address humans' sensory channels, VR applications and devices vary in functionality and purpose (Burdea \u0026amp; Coiffet, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2003\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe history of virtual reality (VR) dates back to the 18th century, beginning with the invention of the first stereoscopic 3D TV by Sir Charles Wheatstone. David Brewster\u0026rsquo;s handheld stereoscope, demonstrated in 1851, paved the way for the development of VR technology, along with Morton Heilmeier\u0026rsquo;s head-mounted display (HMD) and Sensorama. Ivan Sutherland's \u0026ldquo;The Ultimate Display,\u0026rdquo; presented in 1960, was a fundamental blueprint for VR. The early XXI century is known as a \u0026ldquo;VR winter\u0026rdquo;, when VR\u0026rsquo;s potential was explored in depth by government, academy and military research laboratories around the globe (Jerald, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; LaValle, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eVR is a powerful tool for the entertainment industry, especially for gaming and immersive filmmaking (Liu et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; H. Li et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Beyond entertainment, VR has been successfully utilised across industries, including oil and gas, education, medicine, design, therapy, the military, and flight training (Rojas-S\u0026aacute;nchez et al., \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; C, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Utilising VR technologies is cost-effective, as it avoids mistakes before manufacturing, saves time by speeding up iterative processes, provides safe and ecologically valid environments for training, and visualises large datasets that would be incomprehensible with traditional systems (Burdea \u0026amp; Coiffet, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2003\u003c/span\u003e; Joshi et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Soltiyeva et al., \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eExtended Reality, or XR, is a collective term for immersive technologies that combine the physical and digital worlds, encompassing Virtual Reality (VR), Augmented Reality (AR), and Mixed Reality (MR). Augmented Reality (AR) is the concept of digitally overlaying virtual objects onto real-world objects, allowing users to interact with them simultaneously. AR graphics are visible on smartphones, tablets, and other devices, providing a new form of interactive experience for users. Mixed Reality (MR), a combination of AR and VR, blends the physical and digital worlds (Guo et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Morimoto et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Schwaiger et al., \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Research indicates that transforming educational methodologies through VR, AR, and XR technologies provides immersive digital experiences, interactive environments, simulations, and enhanced engagement (Al-Ansi et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eHowever, VR technologies remain in their early stages of development, and there is a gap in their swift implementation and customisation within educational institutions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.3. VR in the education of children with ASD\u003c/h2\u003e \u003cp\u003eImplementing VR technologies in education has several advantages, providing more exploratory, engaging, and playful learning experiences. One of the most important capabilities of VR is the provision of a personalised learning experience and feedback tailored to each student's needs, enabling students to explore VR environments, collaborate with peers worldwide, and make the experience more active and interactive (Marougkas et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). In contrast to traditional learning settings, VR technologies offer safe, cost-effective, and time-saving experiences (Rojas-S\u0026aacute;nchez et al., \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Furthermore, integration of VR technologies in educational settings has shown a positive impact on student learning, academic performance, motivation and engagement. The primary reason for these advantages is that they enable VR students to discover, explore, and interact with objects in ways that are not possible in the physical world (Al-Ansi et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe increasing number of individuals with ASD, along with their growing interest in computerised programs, has prompted numerous studies to develop technologies such as robotics, VR/AR via headsets and goggles, interactive video modelling, and mobile or touchpad devices (Karami et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). VR technologies provide diverse capabilities to meet the needs of both typical users and individuals with special needs (Mesa-Gresa et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Primarily, VR-based training systems can provide a safe, ecologically valid and personalised environment for the individual needs of the ASD population. Because autistic individuals often exhibit differences in peer communication, traditional school settings may present social and environmental demands that can be stressful without appropriate support. In such cases, VR-based learning systems serve as an additional tool for training and educating children with ASD, providing real-life scenarios in a less anxious and more manageable environment (Yuan \u0026amp; Ip, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Zhang et al., \u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). VR technologies, when combined with gamified approaches, can increase motivation, attention, and focus among participants with ASD (Shahab et al., \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Intervention programmes using VR techniques allow repetitive practice by repeatedly presenting tasks and stimuli, an important aspect of interventions for autistic individuals (Didehbani et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Current research covers a wide range of training interventions, including social interaction and communication skills, emotional skills, daily living skills, and cognitive functions (Karami et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.4. Related Work\u003c/h2\u003e \u003cp\u003eEye-tracking technology has long been available for desktop displays; however, in recent years, several commercial solutions have emerged that enable eye tracking on consumer VR head-mounted displays (HMDs). Consequently, research on eye tracking in HMDs has accelerated and expanded significantly (Adhanom et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Learning and training are important applications of VR, and combining it with eye tracking holds great promise for assessing skills and learning outcomes and for improving training (Smutny, \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Adhanom et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Altın et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Devices such as the Meta Quest Pro, HTC Vive Pro Eye, Varjo VR-3 and Pico Neo 3 Pro Eye allow developers to capture gaze direction, fixation points, and pupil indices with high temporal and spatial accuracy (Gonz\u0026aacute;lez \u0026amp; Bozkir, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eRecent developments in computer science have increasingly focused on combining eye-tracking technology with VR to better understand and support social and attentional behaviours in autistic individuals (Chen et al., 2025). One of the earliest examples of this integration was presented by Lahiri et al. (\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2011\u003c/span\u003e), who designed a gaze-sensitive VR system that provided adaptive feedback based on real-time gaze data, demonstrating that eye-tracking could be used effectively to train social interaction skills. Elkin, Zhang, and Reneker (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) developed the VR SAFE project, which features a virtual classroom designed to provide a social context for training activities. Using the HTC Vive Pro Eye VR headset, the researchers tracked participants\u0026rsquo; eye gaze and measured how often they looked into the avatar's eyes. Sanku et al. (\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2023\u003c/span\u003e) developed a VR classroom focused on TEDx-style talks, in which each participant shares personal life incidents. The HMD and wristband collected participants\u0026rsquo; real-time eye-gaze behaviour and heart rate data. Razzak et al. (\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) utilised a VR classroom environment with the HTC Vive Pro Eye headset to manipulate and measure participants\u0026rsquo; attention and engagement. Pino et al. (\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) conducted an eye-tracking\u0026ndash;based comparative study analysing visual attention patterns of children with autism spectrum condition (ASC) when observing real versus virtual avatar faces expressing basic emotions. Using a Tobii eye-tracking system, the authors quantified gaze behaviour by measuring total fixation duration and the number of fixations, and conducted survival analyses of exploration patterns within manually defined areas of interest (eyes and mouth).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eComparison of the related studies\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003e#\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eStudy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eEye-tracking\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eNumber of participants\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSummary and relevance\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDesign of a Gaze-Sensitive Virtual Social Interactive System for Children With Autism (Lahiri et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2011\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eThe Virtual Interactive system with Gaze-sensitive Adaptive Response Technology (VIGART) using VR-based social situation as a platform for delivering individualized feedback based on one\u0026rsquo;s dynamic gaze patterns.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eThe findings indicate that participants' physiological eye responses during VR-based social interactions presented in VIGART may reveal whether they can recognise emotions, consistent with observations from non-VR-based tasks. Thus, there is reason to believe that VIGART could be used as an intervention tool, potentially supporting children with autism spectrum disorder in improving their social skills.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGaze Fixation and Visual Searching Behaviors during an Immersive Virtual Reality Social Skills Training Experience for Children and Youth with Autism Spectrum Disorder: A Pilot Study (Elkin, Zhang \u0026amp; Reneker, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eThe eye tracker on the HTC Vive Pro Eye VR\u003c/p\u003e \u003cp\u003etracked how often the participant looked into the eyes of the avatar, and for how long the participant looked in the virtual environment.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eThe study's findings on gaze fixation and visual search behaviour during simulated training revealed patterns in individuals with mild-to-moderate ASD. The use of this objective measure of gaze fixation behaviour during social skills training may be helpful for similar intervention applications and provide a more accurate measure of individual tendencies in social interaction.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eEnhancing attention in autism spectrum disorder: comparative analysis of virtual reality-based training programs using physiological data(Sanku et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eThe HTC Vive Pro tracks eye information, including pupil size and gaze direction.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eThis study provides a comprehensive evaluation of the Virtual Reality Physiological Data Analysis system, investigating the relationship between physiological data and attention in individuals. Regarding eye-tracking data analysis, the findings provide evidence of improved attention in individuals with ASD.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eUsing virtual reality to enhance attention for autistic spectrum disorder with eye tracking (Razzak et al., \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eThe HTC VIVE Eye Pro headset, eye-tracking, eye openness, pupil size and gaze duration.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eThe findings indicate the effectiveness of VR as\u003c/p\u003e \u003cp\u003ea tool for enhancing attention management in individuals with\u003c/p\u003e \u003cp\u003eASD, highlighting the importance of personalised therapeutic interventions.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eComparing virtual vs real faces expressing emotions in children with autism: An eye-tracking study (Pino et al., \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2021\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTobii T120 Eye Tracker equipment consisting of a GL-2760-LED backlit monitor with a resolution of 1920\u0026times;1080 pixels, which tracks both eyes to an accuracy of 0.5 degrees at a sampling rate of 60 Hz.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eThe results confirm that children with ASD have higher capacities to process and recognize emotions when these are presented by avatar faces.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTogether, these studies demonstrate the promise of VR and eye-tracking technologies for both understanding and supporting individuals with ASD. However, most existing work remains confined to observational or diagnostic applications or to short-term pilot studies. Few systems include real-time gaze analysis with adaptive feedback or establish clear quantitative measures of behavioural improvement. Addressing these limitations, the present study introduces a computationally grounded VR-based training system that leverages real-time eye-tracking and gaze-analysis algorithms to enhance social attention and evaluate measurable training outcomes.\u003c/p\u003e \u003c/div\u003e"},{"header":"3 Concept and Design of the VR System","content":"\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e3.1. Instructional Design Principles\u003c/h2\u003e \u003cp\u003eThe system, called \u003cem\u003eMy Lovely Granny\u0026rsquo;s Farm\u003c/em\u003e, simulates a farm environment containing household objects, animals, and two interactive virtual characters\u0026mdash;a farmer and his son. Users can freely navigate the environment, explore and investigate virtual elements, respond to questions, and complete task-oriented activities. The system integrates real-time eye-tracking to continuously capture users\u0026rsquo; gaze behaviour, including fixation duration, attention shifts between virtual agents, and gaze\u0026ndash;object interactions during task execution. These gaze data are used both for behavioural analysis and to support adaptive interaction within the virtual environment.\u003c/p\u003e \u003cp\u003eGiven that prior research has demonstrated the positive effects of animal-based and nature-inspired environments on the mental health and emotional regulation of individuals with impairments (Berry et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Rodriguez et al., \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; O\u0026rsquo;Haire \u0026amp; Rodriguez, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), the farm setting was intentionally selected for this study. During training, participants interacted with the virtual farmer and the young boy, and their visual attention to socially relevant cues\u0026mdash;such as faces, gestures, and task-related objects\u0026mdash;was objectively monitored via eye tracking. This design enables the assessment of social attention patterns alongside active communication practice. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e2\u003c/span\u003e illustrates the overall system architecture and data flow.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eVirtual farm with all objects, with virtual characters created in Autodesk Maya 2022. The ready scene, including all necessary 3D models and animated characters, was exported to Unity in .fbx format. The Oculus Quest Pro VR headset is used for immersive interaction within virtual environments. The virtual characters, the farmer and son, and the VR farm are demonstrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e3.2. Implementation\u003c/h2\u003e \u003cp\u003eIn this VR-based learning system, characters interact with children using pre-recorded lines. Arm, leg, and torso movements and gestures are included, giving the characters more natural, realistic behaviour and facilitating better user interaction. For more precise interaction with the VR headset, the official Meta libraries were used for camera control and movement rather than the Unity universal libraries. A separate animation accompanies each utterance. Character animations are created in Autodesk Maya and built in Unity using AnimatorController. The AnimatorController uses the State Machine pattern. Animations in this system are considered States, where States are switched through Transitions under specific Conditions. The Conditions vary depending on the Parameters. Figure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003e shows the Component Audio Source.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe option to use the controllers independently has been removed to lower the threshold for children entering VR. It is also possible to move the controllers in space while remaining in place, but this is not recommended for children due to the risk of dizziness. The operator controls the entire session scenario manually via controllers outside the VR zone. Below is an audio source script:\u003c/p\u003e \u003cp\u003ev o i d S t a r t ( )\u003c/p\u003e \u003cp\u003e{\u003c/p\u003e \u003cp\u003eF a r m e r a u d i o\u0026thinsp;=\u0026thinsp;GetComponent\u0026thinsp;\u0026lt;\u0026thinsp;A u d i o S o u r c e \u0026gt; ( ) ;\u003c/p\u003e \u003cp\u003e}\u003c/p\u003e \u003cp\u003ev o i d h e l l o P l a y ( )\u003c/p\u003e \u003cp\u003e{\u003c/p\u003e \u003cp\u003eF a r m e r a u d i o. PlayOneShot ( h e l l o ) ;\u003c/p\u003e \u003cp\u003e}\u003c/p\u003e \u003cp\u003ev o i d h a n d P l a y ( )\u003c/p\u003e \u003cp\u003e{\u003c/p\u003e \u003cp\u003eF a r m e r a u d i o. PlayOneShot ( showhand ) ;\u003c/p\u003e \u003cp\u003e}\u003c/p\u003e \u003cp\u003ev o i d g o o d J o b P l a y ( )\u003c/p\u003e \u003cp\u003e{\u003c/p\u003e \u003cp\u003eF a r m e r a u d i o. PlayOneShot ( g o o d j o b ) ;\u003c/p\u003e \u003cp\u003e}\u003c/p\u003e \u003cp\u003ev o i d i a m H e r e P l a y ( )\u003c/p\u003e \u003cp\u003e{\u003c/p\u003e \u003cp\u003eF a r m e r a u d i o. PlayOneShot ( i a m h e r e ) ;\u003c/p\u003e \u003cp\u003e}\u003c/p\u003e \u003cp\u003eFor more precise interaction with the VR headset, the official Meta libraries were used for camera control and movement rather than the Unity universal libraries.\u003c/p\u003e \u003cp\u003eThe game has two characters, and their animations and replicas are switched through the \u003cem\u003eScenario Controller\u003c/em\u003e script \u0026ndash; the operator \u003cem\u003eButton.One\u003c/em\u003e button switches the index of the required animation for the Farmer character; when the index reaches the end of the list, it wraps around to the beginning. The \u003cem\u003eButton.Two\u003c/em\u003e buttons start the animation of the Farmer's character with the current pointer (index). All of this is performed by the right controller.\u003c/p\u003e \u003cp\u003eThe same logic applies to the boy character (Aidos). Only \u003cem\u003eButton.Three\u003c/em\u003e is used to switch the pointer and \u003cem\u003eButton.Four\u003c/em\u003e on the left controller to start the animation. Controller mapping is demonstrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003e below.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e3.3. Eye Tracking Integration and Logging\u003c/h2\u003e \u003cp\u003eThe Meta Quest Pro has advanced inward-facing infrared (IR) cameras and near-infrared LED illuminators for real-time eye tracking. These cameras capture high-resolution grayscale images of the eyes, allowing the system to detect and monitor pupil position, eye blinks, and gaze direction with high temporal precision (approximately 90\u0026ndash;120 Hz).\u003c/p\u003e \u003cp\u003eThe underlying eye-tracking process combines traditional model-based gaze estimation with data-driven neural network models. The typical stages include:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003ePupil detection using image processing techniques such as ellipse fitting or Hough Circle Transform, or more robustly, via Convolutional Neural Networks (CNNs) trained on large eye image datasets.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003e3D gaze vector estimation using a geometric model of the human eye, calculating the vector from the corneal center through the pupil to determine gaze direction.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eMachine learning-based gaze regression directly maps eye image features to screen coordinates or world-space vectors, thereby increasing robustness to user variability (particularly beneficial for children with Autism Spectrum Disorder).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eThe Meta Quest Pro combines these approaches to achieve enhanced accuracy (\u0026asymp;\u0026thinsp;1\u0026deg;\u0026ndash;2\u0026deg;) and low latency (\u0026lt;\u0026thinsp;50 ms), enabling real-time, gaze-based interaction within immersive virtual environments.\u003c/p\u003e \u003cp\u003eEye-tracking functionality was integrated into the VR application using Meta's Presence Platform SDK in Unity. The following process was implemented:\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eSDK Setup: The Oculus Integration package (version 51 or later) was imported into the Unity project. The EyeTracking feature was enabled in Project Settings\u0026thinsp;\u0026gt;\u0026thinsp;XR Plugin Management.\u003c/p\u003e\u003cp\u003eScripting Access: Eye-tracking data was accessed via Unity C# scripts:\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003ev a r e y e T r a c k i n g D a t a\u0026thinsp;=\u0026thinsp;OVRPlugin. G e t E y e T r a c k i n g S t a t e ( ) ;\u003c/p\u003e \u003cp\u003eV e c t o r 3 g a z e D i r e c t i o n\u0026thinsp;=\u0026thinsp;e y e T r a c k i n g D a t a. E y e G a z e D i r e c t i o n ;\u003c/p\u003e \u003cp\u003eV e c t o r 3 g a z e O r i g i n\u0026thinsp;=\u0026thinsp;e y e T r a c k i n g D a t a. E y e G a z e O r i g i n ;\u003c/p\u003e \u003cp\u003eThese vectors were used to perform ray casts to detect which object the user was looking at.\u003c/p\u003e \u003cp\u003eObject Identification: Each interactive object in the scene (e.g., Farmer, Cow, Tree) was tagged and registered with a unique identifier (name). The object's name was retrieved if the gaze ray intersected a collider.\u003c/p\u003e \u003cp\u003eFixation Duration: A timer was implemented to detect fixations (gaze durations\u0026thinsp;\u0026gt;\u0026thinsp;500 ms), which indicate intentional focus rather than transient glances.\u003c/p\u003e \u003cp\u003eTo capture gaze behavior during training sessions, a custom logging system was developed to store fixation data in JSON format:\u003c/p\u003e \u003cp\u003e{\u003c/p\u003e \u003cp\u003e\u0026rdquo;name \u0026rdquo; : \u0026rdquo; Farmer \u0026rdquo; ,\u003c/p\u003e \u003cp\u003e\u0026rdquo; t i m e \u0026rdquo; : \u0026rdquo; 0, 5 8 \u0026rdquo; ,\u003c/p\u003e \u003cp\u003e\u0026rdquo; d a t e \u0026rdquo; : \u0026rdquo; 2 2. 3 2. 1 0 1 9. 0 2. 2 5 \u0026rdquo;\u003c/p\u003e \u003cp\u003e}\u003c/p\u003e \u003cp\u003eEach entry logs the following:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e\u0026ldquo;name\u0026rdquo;: The name of the object fixated upon;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e\u0026ldquo;time\u0026rdquo;: Duration of the gaze fixation (in seconds);\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e\u0026ldquo;date\u0026rdquo;: Timestamp of the fixation (hh.mm.ss dd.mm.yy).\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003ename = \u0026ldquo;Farmer\u0026rdquo; indicates that the user is fixated on an NPC character labelled \u0026ldquo;Farmer.\u0026rdquo;\u003c/p\u003e \u003cp\u003etime = \u0026ldquo;0,58\u0026rdquo; indicates that the user looked at the object for 0.58 seconds (exceeding the 500 ms threshold).\u003c/p\u003e \u003cp\u003edate = \u0026ldquo;22.32.10 19.02.25\u0026rdquo; provides the time and date of the fixation event (22:32:10 on February 19, 2025).\u003c/p\u003e \u003cp\u003eThe script collects these entries during runtime and writes them to a structured log file at session end or predefined intervals for later analysis of attention patterns, learning behaviour, and interaction frequency.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e6\u003c/span\u003e demonstrates the design and workflow of the VR-based eye-tracking training system.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"4 Methods","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e4.1. Study Design\u003c/h2\u003e \u003cp\u003eThis study employed a within-subject experimental design to explore how a VR system with integrated eye-tracking can support training for individuals with ASD. The goal was to understand how gaze-based interaction can be used to capture patterns of visual attention and enable adaptive system responses during structured VR tasks. Each participant engaged in a predefined set of virtual activities while their gaze behaviour was continuously monitored and recorded. This design enabled consistent observation of user interaction, allowing reliable comparisons while reducing the influence of external factors.\u003c/p\u003e \u003cp\u003eThe eye-tracking component is not in itself a learning intervention. Rather, learning is delivered through structured VR tasks and social interaction scenarios, and eye tracking is used to enable objective observation and analysis. Specifically, the integrated eye tracking system captures gaze behaviour, including fixation duration, gaze transitions, and attention allocation to virtual agents and task-relevant objects. This data is used to identify attention patterns and user interaction behaviour, which serve as measurable indicators of engagement and social attention during training.\u003c/p\u003e \u003cp\u003eThus, gaze tracking in this system plays a dual role: (i) it supports the quantification of user behaviour by providing accurate gaze metrics, and (ii) it provides adaptive system logic by allowing the VR environment to respond to detected attention states when necessary. It is important to note that the behavioural improvements observed in this study are attributable to the VR training design, with gaze tracking serving as a measurement and feedback layer that enhances the system's ability to assess, analyse, and personalise learning.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e4.2 Participants\u003c/h2\u003e \u003cp\u003eThe experimental procedures lasted from February to April 2025, and 10 children aged 3\u0026ndash;6 (8 boys and 2 girls), 10 parents, and 9 specialists from the children's correctional centre participated. The training sessions were conducted at the children\u0026rsquo;s correctional centre \u003cem\u003eDinamica City\u003c/em\u003e, which provides services for children with special educational needs and various developmental diagnoses, including autism, intellectual disability with autistic features, Down syndrome, and others. The centre accommodates children aged 12 months to 17 years, with a maximum capacity of approximately 60. \u003cem\u003eDinamica City\u003c/em\u003e was established in 2021 by Zhunussova Lyazat Kaimbekovna, who serves as both the director and a therapist specialising in ASD. A memorandum of cooperation was signed with the centre, and written consent was obtained from parents permitting their children to participate in the training sessions. Participants received detailed instructions on using the equipment and were introduced to the system's fundamental principles. The primary aim of this stage was to improve the children's communication skills, including eye contact and emotional expression. Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e3\u003c/span\u003e shows participants' characteristics, i.e., age and diagnosis.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e‒ Participants' characteristics\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eID\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAge\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGender\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eDiagnosis\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eASD 1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003emale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAutism, group 4, according to O.S. Nikolskaya's classification\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eASD 2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003efemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAutism, group 2, according to O.S. Nikolskaya's classification\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eASD 3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003emale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eCerebral palsy, moderate intellectual disability with autistic features\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eASD 4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003emale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAutism, group 2, according to O.S. Nikolskaya's classification\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eASD 5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003emale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAtypical autism with intellectual disability\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eASD 6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003emale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAutism, group 4, according to\u003c/p\u003e \u003cp\u003eO.S. Nikolskaya's classification\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eASD 7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003efemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAutism, group 2, according to\u003c/p\u003e \u003cp\u003eO.S. Nikolskaya's classification\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eASD 8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003emale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eDelayed psycho-verbal development, level 2 general speech underdevelopment.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eASD 9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003emale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eCerebral palsy, mild intellectual disability. General speech underdevelopment at level 1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eASD 10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003emale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAutism, group 2, according to\u003c/p\u003e \u003cp\u003eO.S. Nikolskaya's classification\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eAll participants received instructions for using the Meta Quest Pro VR headset and a session on adapting to the virtual environment. They then underwent observational and interaction-based assessments in a structured VR environment prior to exposure to adaptive learning content. To ensure consistency and comfort during the session, participants were allowed a brief familiarisation period within the virtual environment before the assessments. This period enabled them to explore fundamental interactions and navigate the space, thereby reducing potential anxiety or disorientation. During this time, facilitators monitored their responses and provided minimal guidance to encourage independent engagement while ensuring safety and understanding of the virtual setting.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e4.3. Materials\u003c/h2\u003e \u003cp\u003eThe study employed a virtual reality system with an integrated eye-tracking module. Participants interacted with the system via a head-mounted display that captured real-time gaze information, including fixation points and gaze direction. A custom VR environment was developed to present interactive tasks that elicit and sustain visual attention. Within this environment, predefined areas of interest were used to support detailed gaze-based analysis. The VR application was run on a desktop computer, which also recorded interaction data. In addition, dedicated software components were implemented to log gaze coordinates, timestamps, and interaction events for later analysis.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e4.4 Measures\u003c/h2\u003e \u003cp\u003eThe assessment measured time spent interacting (in seconds) with seven in-game virtual agents: \u003cem\u003eFarmer, Boy, Horse, Chicken, Dog, Sheep, and Cow\u003c/em\u003e. Additionally, two key behavioural indicators were recorded: the number of initiated conversations (discrete count) and response time (in seconds). These measures were selected as proxy indicators of engagement (attention duration to virtual agents), social initiation, and information-processing time, which are significant problems in autism and key targets for intervention.\u003c/p\u003e \u003cp\u003eA structured parent-report questionnaire was developed to assess changes in the social functioning of children with ASD before and after the VR training intervention. The questionnaire (APPENDIX A) consisted of 9 social skill items, each rated on a 10-point Likert scale, with 1 indicating minimal proficiency and 10 indicating consistent, independent demonstration of the skill. These metrics included maintaining eye contact, initiating peer conversation, joining group play without prompts, turn-taking, personal hygiene, expressing empathy, recognising facial expressions, responding to greetings, and introducing oneself. Parents provided ratings for each skill before and after the intervention, enabling a paired-comparison analysis. Additionally, a \u0026ldquo;Brief Description\u0026rdquo; field was included for each item to collect qualitative observational data, thereby providing contextual depth. A final open-ended section invited parents to reflect on the effectiveness of the VR system, capturing user perceptions and anecdotal evidence.\u003c/p\u003e \u003cp\u003eThe combination of quantitative behavioural data from the VR environment and qualitative insights from parent reports aimed to provide a multidimensional understanding of the intervention's impact. By triangulating immersive interaction metrics with caregiver observations, the study sought to evaluate not only immediate engagement and responsiveness within the virtual context but also the potential for skill generalisation to real-world social settings. This mixed-methods approach allowed for a more comprehensive evaluation of behavioural change, capturing both measurable improvements and nuanced parental perspectives on developmental progress following the VR training. Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e4\u003c/span\u003e shows eye-tracking and behavioural metrics used to evaluate the VR-based training system.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eOverview of target stimuli and associated gaze- and behaviour-based measurements\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTarget stimulus\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eType of data\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFunction\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFarmer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGaze duration (seconds)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMeasures visual attention directed toward a primary human virtual agent during interaction\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBoy\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGaze duration (seconds)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAssesses attention allocation toward a peer-like virtual character\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHorse\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGaze duration (seconds)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eEvaluates visual attention to non-human animated agents within the environment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eChicken\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGaze duration (seconds)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMeasures engagement with secondary animated objects and agents\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDog\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGaze duration (seconds)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAssesses attention to socially relevant animal agents\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSheep\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGaze duration (seconds)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCaptures the distribution of gaze toward the background animated elements\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGaze duration (seconds)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eEvaluates visual attention to non-human animated agents within the environment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInitiated Conversations\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNumber of conversations\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eQuantifies the frequency of social interaction initiated by the user\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResponse time\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTime duration (seconds)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMeasures the latency of user responses to prompts or social cues\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSocial skills questionnaire (SSQ )\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eScore\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eProvides an external behavioural measure to contextualise in-VR gaze and interaction outcomes\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e4.5. Procedure\u003c/h2\u003e \u003cp\u003eBefore the training sessions began, all participants received safety instructions to ensure a secure and comfortable experience with the VR system. These included guidance on proper use of the VR headset, awareness of the physical boundaries of the training area to prevent collisions or falls, and instructions to notify supervisors immediately if discomfort or dizziness occurs. A trained staff member was present throughout the session to monitor participants and assist as needed. Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e7\u003c/span\u003e demonstrates the floor plan of the training area.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eA parent-reported questionnaire was administered to assess social skills and eye contact in autistic individuals prior to VR-based eye-tracking training. Before the experimental tasks began, the VR head-mounted display was fitted and adjusted for comfort. An eye-tracking calibration procedure was then conducted to align the system with each participant\u0026rsquo;s visual characteristics and ensure accurate gaze measurement. Calibration involved directing the participant\u0026rsquo;s attention to a series of visual targets within the virtual environment and was repeated if necessary to achieve acceptable tracking accuracy. Following calibration, participants completed a set of predefined VR tasks designed to elicit visual attention and interaction with virtual elements. Figure\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003e provides an overview of the study procedure.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eWhen users wear the Oculus Pro headset, they appear in a virtual farm populated with domestic animals, trees, vegetables, and two virtual characters. Users can walk, explore, and interact with virtual objects and the environment. Participants must complete tasks designed to improve their social skills and communication. Depending on the maintenance of eye contact with each virtual agent (e.g., a farmer, a boy, or animals), the system fixes the duration of eye gaze for each participant. Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e5\u003c/span\u003e presents a description of the tasks.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e‒ Description of the tasks\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTask\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQuestions\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFeedback\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eGreeting\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eWhat is your name?\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMy name is Aidos.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHow old are you?\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGreat, I am seven years old\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003eSocial interaction\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHow are you?\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003eI want to show you my grandmother's farm.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHow are you feeling?\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDo you like a horse?\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDo you want to come closer to the horse?\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eDuring task execution, the system continuously recorded gaze behaviour and interaction events. Participants were encouraged to proceed at their own pace, and short breaks were provided when needed to minimise fatigue or discomfort. At the end of the session, the VR equipment was removed, and participants were given time to rest. All recorded data were securely stored for subsequent preprocessing and analysis. After finishing the experimental part, parents filled out post-questionaire to fix changes in social skills and eye-contact.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e4.6. Data analysis\u003c/h2\u003e \u003cp\u003eA quantitative approach was employed to obtain a comprehensive understanding of participants\u0026rsquo; responses to the VR-based training system. To evaluate the effectiveness of the proposed VR-based training system, statistical analyses were selected based on the data structure, sample characteristics, and research objectives. Since gaze-related metrics were collected from the same participants before and after the intervention, paired-sample statistical tests were required to assess within-subject changes attributable to the training. A paired-samples t-test was employed to compare pre- and post-training measurements when the data met the assumptions of approximate normality. The paired \u003cem\u003et\u003c/em\u003e-test is appropriate in this context because it is specifically designed to detect mean differences between two related conditions and provides high statistical power for small-to-moderate sample sizes commonly observed in experimental studies involving children with ASD. Given the potential for non-normal distributions and increased variability in behavioural data, particularly in neurodiverse populations, the Wilcoxon signed-rank test was additionally applied as a non-parametric alternative. This test does not assume normality and is therefore well suited to validating results when distributional assumptions may be violated. The use of both parametric and non-parametric tests strengthens the robustness and reliability of the findings. To complement statistical significance testing, effect size measures (Cohen\u0026rsquo;s \u003cem\u003ed\u003c/em\u003e) were calculated to quantify the magnitude of observed changes. While \u003cem\u003ep\u003c/em\u003e-values indicate whether an effect is statistically detectable, effect sizes provide essential information about the practical and clinical relevance of the intervention, which is especially important in applied VR and autism research. Together, this multi-layered analytical approach ensures that the evaluation captures both statistical reliability and practical significance, offering a comprehensive assessment of the system\u0026rsquo;s impact on gaze behaviour and user engagement.\u003c/p\u003e \u003cp\u003eEngagement was measured using rating scales assessing time-on-task, while the VR system automatically captured eye-tracking data via detailed interaction logs. Each log entry recorded the name of the object or virtual agent the participant focused on (e.g., \u0026ldquo;Farmer\u0026rdquo;), the duration of gaze in seconds (e.g., 0.58 s), and a timestamp indicating the exact time of the interaction. These data allowed for fine-grained analysis of visual attention, including fixation durations, gaze transitions, and interactions with virtual elements. Longer gaze durations were interpreted as higher interest or engagement with the content. For the qualitative analysis, changes in social interaction and emotional responses were assessed using video recordings and field notes collected before and after VR training sessions.\u003c/p\u003e \u003cp\u003eAdditionally, parents completed the Social Skills Questionnaire (SSQ) to provide standardised assessments of participants\u0026rsquo; social skills pre- and post-intervention. Observations from the initial sessions informed iterative improvements to the VR environment, including the introduction of new objects and animals, as well as additional tasks and dialogue prompts designed to enhance engagement and communication skills. This method design enabled triangulation of data sources, combining objective gaze metrics, standardised assessments, and observational insights to provide a robust evaluation of the system\u0026rsquo;s effectiveness in improving attention, engagement, and social behaviour in children with ASD.\u003c/p\u003e \u003c/div\u003e"},{"header":"5 Results","content":"\u003cp\u003eThis study evaluated changes in visual attention toward virtual agents following interaction with the proposed VR-based eye-tracking training system. Gaze behaviour was quantified using fixation duration and gaze engagement metrics derived from the real-time RayCast-based eye-tracking pipeline. Paired-sample \u003cem\u003et\u003c/em\u003e-tests were conducted to compare pre-intervention and post-intervention gaze metrics for each virtual agent. Given the small sample size (\u003cem\u003eN\u003c/em\u003e\u0026thinsp;=\u0026thinsp;11) and potential non-normality of the gaze data, Wilcoxon signed-rank tests were also performed to assess the robustness of the results. Effect sizes were calculated using Cohen\u0026rsquo;s \u003cem\u003ed\u003c/em\u003e to determine the magnitude of the observed changes.\u003c/p\u003e \u003cp\u003eAcross all virtual agents, mean gaze duration increased from the pre-training phase to the post-training phase. At baseline, participants demonstrated limited visual engagement with both social and task-related agents. Following the VR-based training sessions, gaze duration and fixation frequency increased consistently, indicating enhanced visual attention toward the agents embedded in the virtual environment. Although individual variability was observed, the overall trend across participants reflected improved attentional engagement after exposure to the VR system.\u003c/p\u003e \u003cp\u003ePaired-sample \u003cem\u003et\u003c/em\u003e-test results revealed statistically significant increases in gaze duration for all evaluated virtual agents following the intervention (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 for all comparisons). The most substantial statistical effects were observed for agents with guiding or interactive roles (e.g., Farmer and Chicken), indicating that agent function and behavioural salience influenced attention allocation. To confirm these findings, Wilcoxon signed-rank tests were performed, yielding significant pre\u0026ndash;post differences for all agents (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.001), supporting the consistency and reliability of the parametric results. As illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e10\u003c/span\u003e, mean gaze duration increased across all virtual agents following the VR-based training. The most pronounced increases were observed for socially salient agents such as the Farmer and Boy.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eCohen\u0026rsquo;s \u003cem\u003ed\u003c/em\u003e values indicated large effect sizes across all agents, with values ranging from large to very large according to conventional thresholds (\u003cem\u003ed\u003c/em\u003e\u0026thinsp;\u0026ge;\u0026thinsp;0.8), with the most potent effects observed for the Farmer and Chicken agents. The largest effect sizes were associated with socially salient or instruction-oriented agents, while the smallest observed effect still exceeded the threshold for a significant impact. These findings suggest that the VR-based training system produced not only statistically significant but also practically meaningful improvements in visual attention behaviour. Figure\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e11\u003c/span\u003e demonstrates the Cohen\u0026rsquo;s \u003cem\u003ed\u003c/em\u003e effect sizes for changes in gaze duration toward virtual agents following VR-based eye-tracking training.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e12\u003c/span\u003e shows the changes in mean gaze duration before and after training for all virtual agents. A consistent upward trend is observed across all agents, indicating increased visual attention following VR-based gaze-tracking training. While baseline gaze duration was similarly low for all agents, post-training values increased markedly, especially for socially relevant and learning-oriented agents such as Farmer, Boy, and Chicken. These trajectories indicate that the observed improvements are systematic rather than agent-specific anomalies. The uniform direction of change across agents corroborates the statistical evidence and suggests that the intervention resulted in consistent within-subject improvements in visual attention.\u003c/p\u003e \u003cp\u003eThe results demonstrate that the proposed VR-based eye-tracking system consistently improved visual attention across all evaluated virtual agents. Analysis of pre- and post-training gaze data showed a systematic increase in mean gaze duration, as evidenced by statistically significant paired-sample \u003cem\u003et\u003c/em\u003e-tests and confirmed by Wilcoxon signed-rank tests. Effect size analysis indicated large practical effects across agents, demonstrating that the observed changes were robust and meaningful. Together, these findings provide clear evidence that the VR-based training intervention effectively enhanced gaze engagement through immersive, eye-tracking\u0026ndash;driven interaction.\u003c/p\u003e"},{"header":"6 Discussion","content":"\u003cp\u003eThe objective of this study was to investigate whether an immersive VR system with integrated eye-tracking could effectively support visual attention training in individuals with ASD. The results provide strong empirical evidence that the proposed system significantly enhanced gaze engagement across a range of virtual agents. Increases in gaze duration and fixation engagement following the intervention indicate that immersive VR environments can facilitate meaningful improvements in attentional processes in children with ASD. At baseline, participants exhibited limited spontaneous gaze toward virtual agents, which is consistent with previous research documenting atypical attention allocation in ASD. Following VR-based training, participants demonstrated increased and more sustained visual attention, suggesting that repeated interaction within a structured, predictable, and low-anxiety virtual environment promoted attentional regulation. The most significant effects were associated with agents that played explicit social or guiding roles, highlighting the importance of agent design, behavioural clarity, and interaction logic in capturing and maintaining attention within VR-based interventions.\u003c/p\u003e \u003cp\u003eThe findings of this study align with and extend prior work on VR-based interventions for ASD. For example, Lorenzo et al. (\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2016\u003c/span\u003e) demonstrated that immersive VR environments support the development of social understanding and daily living skills by providing structured and repeatable interaction scenarios. Similarly, the significant improvements observed in social engagement metrics and SSQ scores in the present study suggest that immersive VR facilitates safe and effective practice of social behaviours. Li et al. (\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) reported improvements in attention and reduced response times in task-based VR interventions for children with ASD, findings consistent with the decreased response times observed in this study, suggesting more efficient cognitive processing during post-intervention sessions.\u003c/p\u003e \u003cp\u003eFurthermore, the present results complement the systematic review by Mesa-Gresa et al. (\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), which highlighted the potential of immersive, gamified VR applications to improve sensory integration, emotional regulation, and communication skills in individuals with ASD. Improvements in interactions with virtual characters, such as the Farmer, Boy, Chicken, and Dog, suggest increased familiarity, comfort, and engagement in socially meaningful role-play contexts. The substantial improvement in SSQ scores further indicates that immersive VR may positively influence perceived social competence and readiness, consistent with findings reported by Ke and Moon (\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2018\u003c/span\u003e) in avatar-mediated learning environments. Compared with prior studies employing non-immersive or partially immersive systems, the fully immersive VR environment used in this study may account for the more consistent and pronounced improvements across multiple behavioural and cognitive measures. This observation supports the hypothesis that immersion plays a critical role in enhancing engagement and facilitating deeper learning, as suggested by Newbutt et al. (\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). In contrast to studies reporting moderate statistical effects, the proposed system consistently demonstrated strong effects across all key metrics, underscoring the intervention's robustness.\u003c/p\u003e \u003cp\u003eIn contrast to earlier studies using 2D video or static image stimuli (e.g., Chita-Tegmark et al., 2016; Falck-Ytter et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2013\u003c/span\u003e), the present study utilised a fully immersive VR environment featuring seven distinct agents. Unlike prior findings suggesting a preference for non-social stimuli, children in this study exhibited increased gaze duration toward human-like agents, such as the Farmer and Boy, particularly in post-test sessions (Chita-Tegmark et al., 2016; Falck-Ytter et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). This aligns with Chevallier et al. (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2015\u003c/span\u003e), who emphasised the role of realism in promoting social gaze, and Wang et al. (2018), who reported enhanced attention to interactive virtual agents. These results suggest that immersive, context-rich environments may mitigate common attentional biases in ASD and promote more naturalistic social orientation patterns (Chita-Tegmark, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2015\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eBeyond confirming prior findings, this study contributes to the field by integrating high-frequency, real-time eye-tracking into the core interaction logic of an immersive VR system. By combining computational gaze modelling with adaptive task design, the proposed architecture enables attention-responsive training rather than passive assessment. This positions eye-tracking not only as an evaluation tool but as an active mechanism for guiding interaction and learning in VR-based interventions for ASD.\u003c/p\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003e6.1 Limitations of the Study\u003c/h2\u003e \u003cp\u003eWhile this study provides encouraging insights, several limitations should be acknowledged. First, the number of participants was relatively small, which may limit the extent to which the findings can be generalised. Autism spectrum disorder encompasses a wide range of cognitive and behavioural characteristics, and a larger, more diverse sample would likely reveal additional patterns and variations in gaze behaviour that were not captured in this study. Another important consideration relates to the technical constraints of eye-tracking technology. Although the system was designed to operate in real time, its accuracy may be affected by factors such as headset calibration, head motion, lighting conditions, and individual differences in eye physiology. These factors may introduce noise into gaze data and degrade the precision of attention measurements, even when the system operates as intended. The experimental setup itself also introduces certain limitations. The virtual environment was intentionally structured to support controlled observation and consistent interaction; however, this level of control may not fully reflect how users behave in more complex or naturalistic settings. As a result, the extent to which the observed behaviours generalise to real-world contexts remains an open question. In addition, the study primarily relied on gaze-based indicators to infer attention and engagement. While gaze provides valuable insights into visual focus, it represents only one aspect of cognitive and emotional processing. Incorporating additional modalities\u0026mdash;such as facial expression analysis, physiological signals, or performance-based measures\u0026mdash;could offer a more comprehensive understanding of user experience. Finally, the study focused on short-term interactions and did not examine long-term use or learning effects. Future research should explore how user behaviour and engagement evolve over extended periods and whether repeated exposure leads to sustained improvements. Addressing these limitations in future work will help strengthen the robustness and applicability of VR-based eye-tracking systems for individuals with ASD.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003e6.2. Practical implications\u003c/h2\u003e \u003cp\u003eUsing immersive VR technologies to improve an individual's performance expands understanding of personalised education and therapy models. This is consistent with cognitive-behavioural theories that support personalised interventions, reinforcing the idea that different individuals need unique ways of learning. IVRS for ASD children has given rise to new theories of Human-Computer Interaction, investigating the interaction of children with neurodevelopmental disorders with digital environments. The unique ways in which autistic individuals process sensory information have informed more nuanced HCI models, thereby stimulating the development of accessible and engaging interfaces for children with support needs.\u003c/p\u003e \u003cp\u003eUtilising IVRS to simulate different environmental stimuli provides practical insights into sensory processing theories. The findings suggest that controlled sensory exposure in a virtual environment can help children with ASD manage and improve their responses to sensory input, thereby contributing to existing sensory integration and adaptation theories. IVRS contributes to the theory of therapeutic environment design, particularly by elucidating the features that make virtual environments useful for behavioural interventions. The customizable, structured, and predictable nature of VR environments aligns with cognitive-behavioural therapy and deepens theoretical understanding of how environments influence treatment effectiveness.\u003c/p\u003e \u003cp\u003eVR-based learning systems have provided autistic children with a practical, repetitive approach to practising social interactions in a secure, controlled environment. Simulating different social scenarios allows children to practice recognising social cues, understanding appropriate responses, and improving conversational skills. Collaboration among computer scientists, VR developers, educators, and healthcare professionals has led to an interdisciplinary approach to problem-solving in the treatment of ASD. It fosters innovation in software development for neurodiverse populations and leads to the creation of VR applications designed to achieve therapeutic outcomes. These contributions highlight how IVRS advances theoretical models in computer science, education, and therapy and offers scalable, practical solutions that can significantly improve the quality of life of autistic individuals.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003e6.3. Conclusion\u003c/h2\u003e \u003cp\u003eThe application of IVRS for autistic individuals demonstrates substantial potential for enhancing social, cognitive, and behavioural skills, particularly when combined with integrated eye-tracking technologies. The findings of this study show that the proposed IVRS provides an engaging and interactive learning environment that captures and sustains the visual attention of children with ASD more effectively than traditional instructional approaches. By incorporating real-time eye-tracking, the system enables objective measurement of gaze behaviour, fixation patterns, and attention allocation during interaction with virtual agents, offering fine-grained insight into user engagement and learning processes. The multisensory nature of VR, together with gaze-based interaction and feedback, supports sustained attention and focus, thereby enhancing the retention and reinforcement of learned skills. Eye-tracking further enables adaptive system behaviour by allowing the virtual environment to respond dynamically to individual attention patterns, facilitating personalised learning experiences tailored to each child\u0026rsquo;s abilities and sensory preferences. In addition, the controlled and safe nature of VR environments, combined with continuous gaze monitoring, allows children to practise social interactions, interpret social cues, and regulate sensory responses without exposure to real-world risks or overstimulation. Through repeated exposure to gaze-guided social scenarios, children can practise attention to socially relevant stimuli, such as virtual characters and interactive objects, thereby supporting the gradual transfer of these skills to real-life communication contexts. From an empirical perspective, the effectiveness of the proposed eye-tracking\u0026ndash;enabled IVRS was supported by systematic quantitative analysis of pre- and post-intervention gaze metrics. Paired-sample \u003cem\u003et\u003c/em\u003e-tests revealed statistically significant improvements in visual attention across multiple virtual agents, while complementary robustness checks confirmed the consistency of these findings. Effect size analysis indicated large practical effects, demonstrating that the observed changes were not only statistically reliable but also meaningful for user engagement and attention regulation. Collectively, the integration of eye-tracking analytics with immersive VR interaction provides strong evidence that data-driven, gaze-adaptive virtual environments can serve as an effective and scalable intervention framework for supporting social learning in autistic individuals.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e6.4. Future Research Directions\u003c/h2\u003e \u003cp\u003eFuture research will focus on conducting quantitative studies with larger sample sizes to investigate the emotional impact of VR experiences on children with ASD. By collecting and analysing quantitative data, researchers will gain more statistically significant insights into how VR systems affect emotions such as anxiety, joy, or frustration. These studies will use emotion-detection tools, such as facial expression analysis and physiological measures (e.g., heart rate and skin conductance), to objectively assess children's emotional reactions during and after VR sessions.\u003c/p\u003e \u003cp\u003eAnother pivotal area of future research will be the development and implementation of new features in the VR system and the improvement of personalisation. Personalised interactions for each child's unique needs, preferences, and educational needs will be considered. Future systems may include advanced algorithms that automatically adapt scenarios and tasks, or provide real-time feedback based on a child's behaviour, thereby offering a more engaging and personalised experience. This could include customised avatars, customizable environments, and adaptive learning pathways.\u003c/p\u003e \u003cp\u003eOne intended enhancement is the implementation of voice recognition technology to generate more natural and interactive dialogue in VR environments. This will allow children to interact verbally with virtual characters, helping them practice language and social skills in real time. Integrating advanced speech-recognition tools will make interactions more responsive and personalised, potentially improving the communication skills of autistic children.\u003c/p\u003e \u003cp\u003eLastly, future research should investigate how children perceive and respond to changes in the system, including individualised graphic elements tailored to each participant. These elements may include personalised visual stimuli, colours, or themes aligned with a child's interests or sensory preferences. Investigating how children perceive these changes will provide insight into the effectiveness of personalisation and its impact on user engagement, emotional responses, and learning outcomes.\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe sincerely thank all the professionals and researchers who contributed to this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work has been supported by the Academy of Finland Flagship Programme [Grant No. 337653 - Forest-Human-Machine Interplay (UNITE)].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAiganym Soltiyeva: Conceptualisation, Methodology, Formal analysis, Validation, Writing - review \u0026amp; editing.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eWilk Oliveira: Supervision, Writing - review \u0026amp; editing.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAlimanova Madina: Supervision, Writing - review \u0026amp; editing.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eShyngys Adilkhan: Writing - review \u0026amp; editing\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eMarat Urmanov: Writing - review \u0026amp; editing\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThere are no conflicts of interest to declare.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics declarations\u003c/strong\u003e The study was approved by the Research Ethics Committee of SDU University. Written informed consent was obtained from the parents or legal guardians of all participants prior to participation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWritten informed consent was obtained from the parents or legal guardians for the publication of any potentially identifiable images or data included in this article.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and material\u003c/strong\u003e (data transparency)\u003c/p\u003e\n\u003cp\u003eThe original dataset available as supplementary material\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCode availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAdhanom, I. B., MacNeilage, P., \u0026amp; Folmer, E. (2023). Eye Tracking in Virtual Reality: a Broad Review of Applications and Challenges. \u003cem\u003eVirtual Reality\u003c/em\u003e, \u003cem\u003e27\u003c/em\u003e(2), 1481\u0026ndash;1505. https://doi.org/10.1007/s10055-022-00738-z \u003c/li\u003e\n\u003cli\u003eAl-Ansi, A. M., Jaboob, M., Garad, A., \u0026amp; Al-Ansi, A. (2023). Analyzing augmented reality (AR) and virtual reality (VR) recent development in education. Social Sciences \u0026amp; Humanities Open, 8(1), 100532. https://doi.org/10.1016/j.ssaho.2023.100532 \u003c/li\u003e\n\u003cli\u003eAltın, Y., Boşnak, \u0026Ouml;., \u0026amp; Turhan, C. (2025). Examining Virtual Reality Interventions for Social Skills in Children with Autism Spectrum Disorder: A Systematic Review. \u003cem\u003eJournal of Autism and Developmental Disorders\u003c/em\u003e. https://doi.org/10.1007/s10803-025-06741-y \u003c/li\u003e\n\u003cli\u003eAlvari, G., Coviello, L., \u0026amp; Furlanello, C. (2021). EYE-C: Eye-Contact Robust Detection and Analysis during Unconstrained Child-Therapist Interactions in the Clinical Setting of Autism Spectrum Disorders. \u003cem\u003eBrain Sciences\u003c/em\u003e, \u003cem\u003e11\u003c/em\u003e(12), 1555. https://doi.org/10.3390/brainsci11121555 \u003c/li\u003e\n\u003cli\u003eArtiran, S., Ravisankar, R., Luo, S., Chukoskie, L., \u0026amp; Cosman, P. (2022b). Measuring Social Modulation of Gaze in Autism Spectrum Condition With Virtual Reality Interviews. \u003cem\u003eIEEE Transactions on Neural Systems and Rehabilitation Engineering\u003c/em\u003e, \u003cem\u003e30\u003c/em\u003e, 2373\u0026ndash;2384. https://doi.org/10.1109/tnsre.2022.3198933 \u003c/li\u003e\n\u003cli\u003eBurdea, G., \u0026amp; Coiffet, P. (2003). Virtual Reality Technology. \u003cem\u003ePRESENCE Virtual and Augmented Reality\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(6), 663\u0026ndash;664. https://doi.org/10.1162/105474603322955950 \u003c/li\u003e\n\u003cli\u003eBerry, A., Borgi, M., Francia, N., Alleva, E., \u0026amp; Cirulli, F. (2016). Effectiveness of animal-assisted interventions: A systematic review of randomized controlled trials. International Journal of Environmental Research and Public Health, 13(6), Article 618. https://pubmed.ncbi.nlm.nih.gov/24731910/\u003c/li\u003e\n\u003cli\u003eC, A. B. (2024). ENHANCING MILITARY TRAINING THROUGH VR APPLICATIONS. \u003cem\u003eInternational Scientific Journal of Engineering and Management\u003c/em\u003e, \u003cem\u003e03\u003c/em\u003e(05), 1\u0026ndash;9. https://doi.org/10.55041/isjem01739 \u003c/li\u003e\n\u003cli\u003eChevallier C., Parish-Morris J., Tonge N. et al. Susceptibility to distraction in autism spectrum disorder: Probing the integrity of attentional control using eye tracking // Frontiers in Human Neuroscience. \u0026ndash; 2015. \u0026ndash; Vol. 9. \u0026ndash; P. 124-1-124-7. \u003c/li\u003e\n\u003cli\u003eChen, H., \u0026amp; Ren, Y. (2025). Auxiliary diagnostic method for children with autism spectrum disorder based on virtual reality and eye-tracking technology. \u003cem\u003eScientific Reports\u003c/em\u003e, \u003cem\u003e15\u003c/em\u003e(1), 40552. https://doi.org/10.1038/s41598-025-24243-w \u003c/li\u003e\n\u003cli\u003eChita-Tegmark, M. (2015). Social attention in ASD: A review and meta-analysis of eye-tracking studies. \u003cem\u003eResearch in Developmental Disabilities\u003c/em\u003e, \u003cem\u003e48\u003c/em\u003e, 79\u0026ndash;93. https://doi.org/10.1016/j.ridd.2015.10.011 \u003c/li\u003e\n\u003cli\u003eChoi, H., \u0026amp; Nam, S. (2025). Effect of eye-tracking-based depth perception visualization interface on virtual reality user experience. \u003cem\u003eVirtual Reality\u003c/em\u003e, \u003cem\u003e30\u003c/em\u003e(1). https://doi.org/10.1007/s10055-025-01260-8 \u003c/li\u003e\n\u003cli\u003eCreed, C., Al-Kalbani, M., Theil, A., Sarcar, S., \u0026amp; Williams, I. (2023). Inclusive AR/VR: accessibility barriers for immersive technologies. \u003cem\u003eUniversal Access in the Information Society\u003c/em\u003e, \u003cem\u003e23\u003c/em\u003e(1), 59\u0026ndash;73. https://doi.org/10.1007/s10209-023-00969-0 \u003c/li\u003e\n\u003cli\u003e\u003cem\u003eData and statistics on autism spectrum disorder\u003c/em\u003e. (2025, May 27). Autism Spectrum Disorder (ASD). https://www.cdc.gov/autism/data-research/index.html \u003c/li\u003e\n\u003cli\u003eDidehbani, N., Allen, T., Kandalaft, M., Krawczyk, D., \u0026amp; Chapman, S. (2016). Virtual Reality Social Cognition Training for children with high functioning autism. \u003cem\u003eComputers in Human Behavior\u003c/em\u003e, \u003cem\u003e62\u003c/em\u003e, 703\u0026ndash;711. https://doi.org/10.1016/j.chb.2016.04.033 \u003c/li\u003e\n\u003cli\u003eDudley, J., Yin, L., Garaj, V., \u0026amp; Kristensson, P. O. (2023). Inclusive Immersion: a review of efforts to improve accessibility in virtual reality, augmented reality and the metaverse. \u003cem\u003eVirtual Reality\u003c/em\u003e, \u003cem\u003e27\u003c/em\u003e(4), 2989\u0026ndash;3020. https://doi.org/10.1007/s10055-023-00850-8 \u003c/li\u003e\n\u003cli\u003eElkin, T. D., Zhang, Y., \u0026amp; Reneker, J. C. (2022). Gaze Fixation and Visual Searching Behaviors during an Immersive Virtual Reality Social Skills Training Experience for Children and Youth with Autism Spectrum Disorder: A Pilot Study. \u003cem\u003eBrain Sciences\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(11), 1568. https://doi.org/10.3390/brainsci12111568 \u003c/li\u003e\n\u003cli\u003eFalck-Ytter T., B\u0026ouml;lte S., Gredeb\u0026auml;ck G. Eye tracking in early autism research // J of Neurodevelopmental Disorders. \u0026ndash; 2013. \u0026ndash; Vol. 5, Issue 1. \u0026ndash; P. 28-1-28-13. \u003c/li\u003e\n\u003cli\u003eGonz\u0026aacute;lez, N. M., \u0026amp; Bozkir, E. (2024). Eye-Tracking Devices for Virtual and Augmented Reality Metaverse Environments and Their Compatibility with the European Union General Data Protection Regulation. \u003cem\u003eDigital Society\u003c/em\u003e, \u003cem\u003e3\u003c/em\u003e(2). https://doi.org/10.1007/s44206-024-00128-9 \u003c/li\u003e\n\u003cli\u003eGrasnick, A. (2021). \u003cem\u003eBasics of virtual reality\u003c/em\u003e. https://doi.org/10.1007/978-3-662-64201-6 \u003c/li\u003e\n\u003cli\u003eGuo, X., Guo, Y., \u0026amp; Liu, Y. (2021). The Development of Extended Reality in Education: Inspiration from the Research Literature. \u003cem\u003eSustainability\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e(24), 13776. https://doi.org/10.3390/su132413776 \u003c/li\u003e\n\u003cli\u003eJerald, J. (2015). \u003cem\u003eThe VR Book: Human-Centered Design for Virtual Reality\u003c/em\u003e. https://doi.org/10.1145/2792790 \u003c/li\u003e\n\u003cli\u003eJeyarani, R. A., \u0026amp; Senthilkumar, R. (2023). Eye Tracking Biomarkers for Autism Spectrum Disorder Detection using Machine Learning and Deep Learning Techniques: Review. \u003cem\u003eResearch in Autism Spectrum Disorders\u003c/em\u003e, \u003cem\u003e108\u003c/em\u003e, 102228. https://doi.org/10.1016/j.rasd.2023.102228 \u003c/li\u003e\n\u003cli\u003eJoshi, S., Hamilton, M., Warren, R., Faucett, D., Tian, W., Wang, Y., \u0026amp; Ma, J. (2020). Implementing Virtual Reality technology for safety training in the precast/ prestressed concrete industry. \u003cem\u003eApplied Ergonomics\u003c/em\u003e, \u003cem\u003e90\u003c/em\u003e, 103286. https://doi.org/10.1016/j.apergo.2020.103286 \u003c/li\u003e\n\u003cli\u003eKarami, B., Koushki, R., Arabgol, F., Rahmani, M., \u0026amp; Vahabie, A. (2021). Effectiveness of Virtual/Augmented Reality\u0026ndash;Based therapeutic Interventions on Individuals with Autism Spectrum Disorder: A Comprehensive Meta-Analysis. \u003cem\u003eFrontiers in Psychiatry\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e. https://doi.org/10.3389/fpsyt.2021.665326 \u003c/li\u003e\n\u003cli\u003eKaur, B., \u0026amp; Josan, G. S. (2025). Bringing Faces to Life: A survey on realistic facial expressions in 3D virtual characters. \u003cem\u003eInternational Journal of Human-Computer Interaction\u003c/em\u003e, 1\u0026ndash;42. https://doi.org/10.1080/10447318.2025.2561766 \u003c/li\u003e\n\u003cli\u003eKe, F., \u0026amp; Moon, J. (2018). Virtual collaborative gaming as social skills training for high‐functioning autistic children. \u003cem\u003eBritish Journal of Educational Technology\u003c/em\u003e, \u003cem\u003e49\u003c/em\u003e(4), 728\u0026ndash;741. https://doi.org/10.1111/bjet.12626 \u003c/li\u003e\n\u003cli\u003eKoumpouros, Y. (2025). Digital Horizons: Enhancing Autism Support with Augmented Reality. \u003cem\u003eJournal of Autism and Developmental Disorders\u003c/em\u003e. https://doi.org/10.1007/s10803-024-06709-4 \u003c/li\u003e\n\u003cli\u003eLahiri, U., Warren, Z., \u0026amp; Sarkar, N. (2011). Design of a Gaze-Sensitive Virtual Social interactive system for children with autism. \u003cem\u003eIEEE Transactions on Neural Systems and Rehabilitation Engineering\u003c/em\u003e, \u003cem\u003e19\u003c/em\u003e(4), 443\u0026ndash;452. https://doi.org/10.1109/tnsre.2011.2153874 \u003c/li\u003e\n\u003cli\u003eLaValle, S. M. (2023). \u003cem\u003eVirtual reality\u003c/em\u003e. Cambridge university press.\u003c/li\u003e\n\u003cli\u003eLi, H., Lo, C., Smith, A., \u0026amp; Yu, Z. (2022). The development of virtual production in film industry in the past decade. In \u003cem\u003eLecture notes in computer science\u003c/em\u003e (pp. 221\u0026ndash;239). https://doi.org/10.1007/978-3-031-06047-2_16 \u003c/li\u003e\n\u003cli\u003eLi, N., Tian, M., Yang, Y., Liu, Z., Sun, L., \u0026amp; Li, B. (2025). The effect of fully immersive virtual reality technology combined with psychological and behavioral intervention on autism spectrum disorder. \u003cem\u003eBMC Psychology\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e(1), 1120. https://doi.org/10.1186/s40359-025-03460-y \u003c/li\u003e\n\u003cli\u003eLiu, Z., Jin, Y., Ma, M., \u0026amp; Li, J. (2022). A comparison of Immersive and Non-Immersive VR for the education of filmmaking. \u003cem\u003eInternational Journal of Human-Computer Interaction\u003c/em\u003e, \u003cem\u003e39\u003c/em\u003e(12), 2478\u0026ndash;2491. https://doi.org/10.1080/10447318.2022.2078462 \u003c/li\u003e\n\u003cli\u003eLorenzo, G., Lled\u0026oacute;, A., Pomares, J., \u0026amp; Roig, R. (2016). Design and application of an immersive virtual reality system to enhance emotional skills for children with autism spectrum disorders. \u003cem\u003eComputers \u0026amp; Education\u003c/em\u003e, \u003cem\u003e98\u003c/em\u003e, 192\u0026ndash;205. https://doi.org/10.1016/j.compedu.2016.03.018 \u003c/li\u003e\n\u003cli\u003eLorenzo, G. G., Newbutt, N. N., \u0026amp; Lorenzo-Lled\u0026oacute;, A. A. (2023). Designing virtual reality tools for students with Autism Spectrum Disorder: A systematic review. \u003cem\u003eEducation and Information Technologies\u003c/em\u003e, \u003cem\u003e28\u003c/em\u003e(8), 9557\u0026ndash;9605. https://doi.org/10.1007/s10639-022-11545-z \u003c/li\u003e\n\u003cli\u003eMarougkas, A., Troussas, C., Krouska, A., \u0026amp; Sgouropoulou, C. (2023). Virtual Reality in Education: A review of learning theories, approaches and methodologies for the last decade. \u003cem\u003eElectronics\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(13), 2832. https://doi.org/10.3390/electronics12132832 \u003c/li\u003e\n\u003cli\u003eMesa-Gresa, P., Gil-G\u0026oacute;mez, H., Lozano-Quilis, J., \u0026amp; Gil-G\u0026oacute;mez, J. (2018). Effectiveness of Virtual Reality for Children and Adolescents with Autism Spectrum Disorder: An Evidence-Based Systematic Review. \u003cem\u003eSensors\u003c/em\u003e, \u003cem\u003e18\u003c/em\u003e(8), 2486. https://doi.org/10.3390/s18082486 \u003c/li\u003e\n\u003cli\u003eMorimoto, T., Kobayashi, T., Hirata, H., Otani, K., Sugimoto, M., Tsukamoto, M., Yoshihara, T., Ueno, M., \u0026amp; Mawatari, M. (2022). XR (Extended Reality: Virtual Reality, Augmented Reality, Mixed Reality) Technology in Spine Medicine: status quo and Quo vadis. \u003cem\u003eJournal of Clinical Medicine\u003c/em\u003e, \u003cem\u003e11\u003c/em\u003e(2), 470. https://doi.org/10.3390/jcm11020470 \u003c/li\u003e\n\u003cli\u003eMeta for Developers. (2025, November 10). Get started with basic controller input in Unity. Meta. https://developers.meta.com/horizon/documentation/unity/unity-tutorial-basic-controller-input/\u003c/li\u003e\n\u003cli\u003eNewbutt, N., Bradley, R., \u0026amp; Conley, I. (2019). Using Virtual Reality Head-Mounted Displays in Schools with Autistic Children: Views, Experiences, and Future Directions. \u003cem\u003eCyberpsychology Behavior and Social Networking\u003c/em\u003e, \u003cem\u003e23\u003c/em\u003e(1), 23\u0026ndash;33. https://doi.org/10.1089/cyber.2019.0206 \u003c/li\u003e\n\u003cli\u003eO\u0026rsquo;Haire, M. E., \u0026amp; Rodriguez, K. E. (2021). Animal-assisted activities improve social behaviors in psychiatrically hospitalized youth with autism. Journal of Applied Developmental Psychology, 72, Article 101272. https://pubmed.ncbi.nlm.nih.gov/30818971/\u003c/li\u003e\n\u003cli\u003ePartarakis, N., \u0026amp; Zabulis, X. (2024). A review of immersive Technologies, knowledge representation, and AI for Human-Centered Digital Experiences. \u003cem\u003eElectronics\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e(2), 269. https://doi.org/10.3390/electronics13020269 \u003c/li\u003e\n\u003cli\u003ePino, M. C., Vagnetti, R., Valenti, M., \u0026amp; Mazza, M. (2021). Comparing virtual vs real faces expressing emotions in children with autism: An eye-tracking study. \u003cem\u003eEducation and Information Technologies\u003c/em\u003e, \u003cem\u003e26\u003c/em\u003e(5), 5717\u0026ndash;5732. https://doi.org/10.1007/s10639-021-10552-w \u003c/li\u003e\n\u003cli\u003eRazzak, R., Li, Y., He, J., Jung, S., Mei, C., \u0026amp; Huang, Y. (2024). Using virtual reality to enhance attention for autistic spectrum disorder with eye tracking. \u003cem\u003eHigh-Confidence Computing\u003c/em\u003e, \u003cem\u003e5\u003c/em\u003e(1), 100234. https://doi.org/10.1016/j.hcc.2024.100234 \u003c/li\u003e\n\u003cli\u003eRodriguez, K. E., Greer, J., Yatcilla, J. K., Beck, A. M., \u0026amp; O\u0026rsquo;Haire, M. E. (2019). The effects of assistance dogs on psychosocial health and wellbeing: A systematic literature review. PLOS ONE, 14(8), e0256071. https://doi.org/10.1371/journal.pone.0256071\u003c/li\u003e\n\u003cli\u003eRojas-S\u0026aacute;nchez, M. A., Palos-S\u0026aacute;nchez, P. R., \u0026amp; Folgado-Fern\u0026aacute;ndez, J. A. (2022). Systematic literature review and bibliometric analysis on virtual reality and education. \u003cem\u003eEducation and Information Technologies\u003c/em\u003e, \u003cem\u003e28\u003c/em\u003e(1), 155\u0026ndash;192. https://doi.org/10.1007/s10639-022-11167-5 \u003c/li\u003e\n\u003cli\u003eSanku, B. S., Li, Y., Jung, S., Mei, C., \u0026amp; He, J. (2023). Enhancing attention in autism spectrum disorder: comparative analysis of virtual reality-based training programs using physiological data. \u003cem\u003eFrontiers in Computer Science\u003c/em\u003e, \u003cem\u003e5\u003c/em\u003e. https://doi.org/10.3389/fcomp.2023.1250652 \u003c/li\u003e\n\u003cli\u003eShahab, M., Taheri, A., Mokhtari, M., Shariati, A., Heidari, R., Meghdari, A., \u0026amp; Alemi, M. (2021). Utilizing social virtual reality robot (V2R) for music education to children with high-functioning autism. \u003cem\u003eEducation and Information Technologies\u003c/em\u003e, \u003cem\u003e27\u003c/em\u003e(1), 819\u0026ndash;843. https://doi.org/10.1007/s10639-020-10392-0 \u003c/li\u003e\n\u003cli\u003eSchloss, I., Dillon, A., \u0026amp; Bailey, J. (2025). Why use virtual reality as a treatment for children and youth\u0026rsquo;s mental health and wellbeing: a review. \u003cem\u003eDigital Society\u003c/em\u003e, \u003cem\u003e4\u003c/em\u003e(2). https://doi.org/10.1007/s44206-025-00193-8 \u003c/li\u003e\n\u003cli\u003eSchwaiger, M., Krajnčan, M., Vukovič, M., Jenko, M., \u0026amp; Doz, D. (2024). Educators\u0026rsquo; opinions about VR/AR/XR: An exploratory study. \u003cem\u003eEducation and Information Technologies\u003c/em\u003e. https://doi.org/10.1007/s10639-024-12808-7 \u003c/li\u003e\n\u003cli\u003eSilva, R. M., Martins, P., \u0026amp; Rocha, T. (2024). Virtual reality educational scenarios for students with ASD: Instruments validation and design of STEM programmatic contents. \u003cem\u003eResearch in Autism Spectrum Disorders\u003c/em\u003e, \u003cem\u003e119\u003c/em\u003e, 102521. https://doi.org/10.1016/j.rasd.2024.102521 \u003c/li\u003e\n\u003cli\u003eSmutny, P. (2022). Learning with virtual reality: a market analysis of educational and training applications. \u003cem\u003eInteractive Learning Environments\u003c/em\u003e, \u003cem\u003e31\u003c/em\u003e(10), 6133\u0026ndash;6146. https://doi.org/10.1080/10494820.2022.2028856 \u003c/li\u003e\n\u003cli\u003eSoltiyeva, A., Oliveira, W., Madina, A., Adilkhan, S., Urmanov, M., \u0026amp; Hamari, J. (2023). My Lovely Granny\u0026rsquo;s Farm: An immersive virtual reality training system for children with autism spectrum disorder. \u003cem\u003eEducation and Information Technologies\u003c/em\u003e, \u003cem\u003e28\u003c/em\u003e(12), 16887\u0026ndash;16907. https://doi.org/10.1007/s10639-023-11862-x \u003c/li\u003e\n\u003cli\u003eSorrentino, G., Edwards, M. G., Baldini, N., Mustile, M., Lejeune, T., \u0026amp; Everard, G. (2025). REAsmash-ET: a methodological framework for combined cognitive and motor assessment through eye-tracking and kinematic metrics in immersive VR search-and-reach task. \u003cem\u003eJournal of NeuroEngineering and Rehabilitation\u003c/em\u003e, \u003cem\u003e23\u003c/em\u003e(1), 25. https://doi.org/10.1186/s12984-025-01844-0 \u003c/li\u003e\n\u003cli\u003eStephenson, L. J., Edwards, S. G., \u0026amp; Bayliss, A. P. (2021). From gaze perception to social cognition: the Shared-Attention System. \u003cem\u003ePerspectives on Psychological Science\u003c/em\u003e, \u003cem\u003e16\u003c/em\u003e(3), 553\u0026ndash;576. https://doi.org/10.1177/1745691620953773 \u003c/li\u003e\n\u003cli\u003eStuart, N., Whitehouse, A., Palermo, R., Bothe, E., \u0026amp; Badcock, N. (2022). Eye gaze in Autism Spectrum Disorder: A review of Neural Evidence for the eye Avoidance hypothesis. \u003cem\u003eJournal of Autism and Developmental Disorders\u003c/em\u003e, \u003cem\u003e53\u003c/em\u003e(5), 1884\u0026ndash;1905. https://doi.org/10.1007/s10803-022-05443-z \u003c/li\u003e\n\u003cli\u003eWang, L., Wang, B., Wu, C., Wang, J., \u0026amp; Sun, M. (2023). Autism spectrum Disorder: neurodevelopmental risk factors, biological mechanism, and precision therapy. \u003cem\u003eInternational Journal of Molecular Sciences\u003c/em\u003e, \u003cem\u003e24\u003c/em\u003e(3), 1819. https://doi.org/10.3390/ijms24031819 \u003c/li\u003e\n\u003cli\u003eYang, X., Wu, J., Ma, Y., Yu, J., Cao, H., Zeng, A., Fu, R., Tang, Y., \u0026amp; Ren, Z. (2024). Effectiveness of virtual reality technology intervention in improving social skills of children and adolescents with autism:A systematic review (Preprint). \u003cem\u003eJournal of Medical Internet Research\u003c/em\u003e. https://doi.org/10.2196/60845 \u003c/li\u003e\n\u003cli\u003eYuan, S. N. V., \u0026amp; Ip, H. H. S. (2018). Using virtual reality to train emotional and social skills in children with autism spectrum disorder. \u003cem\u003eLondon Journal of Primary Care\u003c/em\u003e, \u003cem\u003e10\u003c/em\u003e(4), 110\u0026ndash;112. https://doi.org/10.1080/17571472.2018.1483000 \u003c/li\u003e\n\u003cli\u003eYu, C., Wang, S., Zhang, D., Zhang, Y., Cen, C., You, Z., Zou, X., Deng, H., \u0026amp; Li, M. (2023, October 20). \u003cem\u003eHSVRS: a virtual reality system of the Hide-and-Seek game to enhance gaze fixation ability for autistic children\u003c/em\u003e. arXiv.org. https://arxiv.org/abs/2310.13482 \u003c/li\u003e\n\u003cli\u003eZhang, M., Ding, H., Naumceska, M., \u0026amp; Zhang, Y. (2022). Virtual Reality Technology as an Educational and Intervention Tool for Children with Autism Spectrum Disorder: Current Perspectives and Future Directions. \u003cem\u003eBehavioral Sciences\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(5), 138. https://doi.org/10.3390/bs12050138 \u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Virtual Reality, Immersive technologies, eye tracking, Autism Spectrum Disorders, assistive technologies","lastPublishedDoi":"10.21203/rs.3.rs-8852358/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8852358/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eEffective training of gaze-related behaviours in Autism Spectrum Disorder (ASD) requires systems capable of capturing and responding to visual attention with high temporal precision. However, existing virtual reality (VR) applications lack architectures for real-time, fine-grained gaze analysis, limiting their usefulness for adaptive training scenarios. To address this problem, this article presents the design, implementation, and evaluation of a VR-based eye-tracking system that enables fine-grained, real-time analysis of visual attention during interactions with virtual agents. The system incorporates a high-frequency gaze-event detection pipeline that uses RayCast-based sampling at 0.02-second intervals, a modular logging architecture for capturing temporal gaze metrics, and an adaptive task framework that dynamically responds to user attention. Implemented on the Oculus Quest Pro, the system records fixation distribution, gaze-to-agent transitions, and object-interaction patterns with millisecond-level precision. An experimental study with children with ASD demonstrates the system\u0026rsquo;s capability to accurately track gaze behaviour across multiple virtual characters and interactive elements. The results show the feasibility of using computational gaze modelling to support training-oriented VR environments for ASD. This work contributes a VR architecture for real-time gaze analysis and establishes a technical foundation for future intelligent, gaze-adaptive VR interventions.\u003c/p\u003e","manuscriptTitle":"A Virtual Reality-Based Eye-Tracking System for Training Autistic Children","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-26 15:45:55","doi":"10.21203/rs.3.rs-8852358/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"69c5bc87-6a0a-4548-b331-acbd17b152a6","owner":[],"postedDate":"April 26th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-26T15:45:56+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-26 15:45:55","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8852358","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8852358","identity":"rs-8852358","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}