Autism spectrum disorder (ASD) is a neurodevelopmental condition characterized by persistent difficulties in social communication and interaction, as well as restricted and repetitive patterns of behavior [1]. Symptoms typically begin to emerge within the first 2 years of life [2]. Because ASD manifests early in development and exerts cumulative effects across a child’s social, linguistic, and cognitive developmental trajectories, early diagnosis and screening play a critical role in improving treatment outcomes and long-term prognosis [3,4]. Globally, the prevalence of ASD has continued to rise, posing a significant burden on public health and educational systems [5]. Evidence from Korean birth cohort studies similarly indicates an increase in diagnostic rates, suggesting that this trend is not unique to Western countries [6]. This growing prevalence further underscores the urgent need to establish effective systems for early ASD screening and intervention.

In the context of ASD, early screening and diagnosis serve distinct but complementary purposes. Screening is a quick and simple procedure aimed at detecting developmental delays or early signs of autism, helping to determine whether a child is following an expected developmental trajectory or requires more in-depth professional evaluation [7]. At the diagnostic stage, a more comprehensive and detailed assessment is conducted by trained specialists. Importantly, while a screening result does not constitute a diagnosis, a positive screening outcome should lead to a thorough diagnostic evaluation, thereby serving as a critical gateway to early intervention and support services [8]. Early screening not only facilitates earlier diagnosis but also plays a key role in reducing children’s frustration and improving their quality of life by enabling timely access to intervention [9]. Prior research has demonstrated that therapeutic interventions initiated between 18 and 30 months of age can lead to significant improvements in intellectual functioning and adaptive skills, as well as diagnostic outcomes 2 years later [10]. In a randomized controlled trial of children showing early signs of ASD, those who received early intervention exhibited significantly fewer ASD-related behaviors and were less likely to meet diagnostic criteria for ASD at age 3 (7%) compared to those receiving usual care (21%) [11]. These findings indicate that the earlier intervention is initiated, the greater the improvement in symptoms and the more favorable the developmental prognosis [12,13]. In line with this evidence, the American Academy of Pediatrics recommends autism-specific screening at 18 and 24 months [14], and the implementation of standardized screening procedures has been shown to reduce disparities in service access and increase early diagnosis rates [15].

Commonly used screening tools for ASD include the Social Responsiveness Scale 2nd edition (SRS-2) [16,17], the Social Communication Questionnaire (SCQ) [18,19], and the Behavior Development Screening for Toddlers (BeDevel), which was developed in Korea [20,21]. BeDevel is specifically designed for children aged 9-42 months and was developed to reflect the Korean cultural context. It consists of a parent interview (BeDevel-Interview, BeDevel-I) and a play observation (BeDevel-Play, BeDevel-P) and has demonstrated high reliability and validity. However, questionnaire-based assessments rely heavily on the perceptions and experiences of informants, which may limit the objectivity of the evaluation [22]. In addition, most questionnaires use total scores as the basis for screening, which carries the risk of overlooking important diagnostic characteristics [23,24]. Assessments conducted in clinical or specialized settings also require examiners with professional training and involve time- and resource-intensive procedures, which can pose additional barriers to access [25]. Moreover, for young children, observed behaviors may vary considerably depending on where and with whom the assessment is conducted. Behaviors observed during brief interactions with unfamiliar adults in an unfamiliar environment may not accurately reflect a child’s everyday behavioral patterns. For this reason, observation in more familiar and naturalistic environments is often essential for accurate diagnostic evaluation, though not always easy to implement in practice. Therefore, assessments conducted in settings less affected by environmental factors may be particularly useful for initial screening.

Despite the continuous increase in the prevalence of ASD, the Korean healthcare system faces challenges such as low referral rates to child psychiatrists, which make timely screening and diagnosis difficult [26,27]. As a result, it is challenging to connect children with appropriate support systems, and most psychosocial treatments and services are not covered by the national health insurance system [28]. These challenges could be alleviated to some extent if easier and more objective ASD screening tools were available. Developing a simple and objective screening tool could greatly expand the opportunities for early detection and intervention, as it would facilitate early screening and increase access to necessary services. To achieve this, a technology-based approach, such as a video-based presentation method, that delivers the same stimuli under identical conditions could be considered. Such an approach would offer the advantages of simplifying the traditionally complex clinical evaluation process, reducing the need for professional training, and enabling large-scale screenings [29]. There have also been studies that analyze the reactions of children watching specific videos. Most of these studies were conducted in children who had already been diagnosed with ASD [30]. In addition, most studies have only compared the severity of ASD behavior with the response of children watching videos [31]. In other words, studies aimed at diagnosing or screening ASD were rare. Further, the items of the response to be evaluated were limited to one or two levels per video, and there were few attempts to quantitatively evaluate by setting criteria for this [32].

Therefore, we developed a 5-minute scenario-based video to evaluate children’s behavioral responses while watching the video and to conduct an ASD screening assessment. The video was designed to include activities that elicit key social-communicative behaviors such as joint attention, response to name, social referencing, eye contact, imitation, social smiling, and pointing. The composition and structure of the scenario were developed with reference to the BeDevel-P. Evaluating ASD through responses to time-structured scenarios allows for reduced examiner intervention, thereby minimizing variability in results and potential sources of bias. Furthermore, the test can be administered in various environments where videos can be played, enabling repeated assessments without restrictions of space or time. In addition, because multiple behavioral domains can be assessed within a single scenario, the tool facilitates a more comprehensive yet efficient evaluation while retaining the advantages of a screening measure. Taken together, these considerations highlight the need for a simple, objective, and standardized approach to early ASD screening. Therefore, the present study aimed to develop and validate a scenario-based ASD early screening tool and to explore its potential utility for early identification and intervention.

The analysis of the differences in responses between the ASD and non-ASD groups revealed significant differences in response to name, eye contact, pointing behavior, initiation of joint attention, social referencing, and social smiling. When compared to previous studies, such as the research using the BeDevel-P, which involves direct face-to-face interaction with the examiner and found group differences in most items [20,21], our study showed differences in some activities but did not find group differences in response to joint attention and imitation behavior. However, regarding response to joint attention, our findings are like previous research that suggests children older than 24 months, regardless of whether they are diagnosed with ASD, typically show appropriate responses, thus limiting the discriminative power of this behavior [39,40]. Ad-ditionally, the lack of significant differences in imitation behavior can be attributed to the video-based method used to assess this behavior. Considering that imitation behavior involves interest in others’ actions and attention to the person performing those actions [41,42], the use of video rather than live interaction may have influenced the ability to accurately assess imitation behavior.

When examining individual behaviors, particularly response to name and pointing behavior, the screening tool showed high discrimination power. This suggests that the scenario-based ASD early screening tool could be useful for assessing responsiveness in young children, as these behaviors involve direct responses to stimuli. Although the artificial intelligence (AI)-based content screening tool can elicit responses from children and trigger simple social or spontaneous behaviors, it presents practical challenges in evaluating interpersonal interactions [43,44]. On the other hand, the tool’s use of video-based structured situations to present clear response characteristics is expected to be beneficial for identifying these behaviors [45]. Additionally, significant differences were found in eye contact, social referencing, and social smiling, further supporting the tool’s usefulness for assessing responsiveness. Although the significance level for initiation of joint attention was lower, group differences were still observed, suggesting that the tool may be useful for prompting spontaneous behaviors. This indicates that future research could explore the potential of applying this tool to individual behaviors, considering the differences in discriminatory power across activities, and potentially assigning weights to each activity. This approach could be seen as the BeDevel-P, where the same target behaviors (e.g., social referencing) are assessed in multiple contexts (e.g., block play, hiding play) [20,21].

To effectively implement the scenario-based ASD early screening tool, clear and practical criteria, like those used in existing screening tools, are necessary [46]. Using the two scoring methods, statistically significant differences between groups were observed. The ability to clearly identify these differences enhances the potential utility of the tool as a screening tool. In addition, the screening system that presents structured, video-based activities consistently provides a series of standardized tasks designed to elicit children’s behaviors. This approach increases the consistency of their responses, minimizes examiner bias, and reduces the influence of subjective judgment. However, when examining sensitivity and specificity through the receiver operating characteristic curve, the scenario-based ASD early screening tool demonstrated high sensitivity, indicating that it could effectively identify ASD. However, its lower specificity suggests a potential for overidentification, leading to false positives. This is particularly relevant for early childhood development, as various psychological, social, and physiological factors influence development, and failure to account for these variables could impact the ability to accurately assess ASD traits in children [47]. Additionally, the environment in which the tool is administered, along with the child’s health status and mood at the time of assessment, could influence results [48]. This may explain the challenges in achieving higher specificity. Therefore, in practical applications, it may be beneficial to assess children in environments where they feel comfortable, based on their everyday experiences. Furthermore, it is crucial to clearly define the purpose of the tool’s use and establish specific guidelines for utilizing it, especially for primary screening purposes.

Considering the strengths of a scenario-based ASD early screening tool, several potential applications can be anticipated. First, its use in the home environment can be considered. Since the home is the everyday setting where young children spend most of their time, it allows for assessing behavioral characteristics when children are in their most comfortable state [49]. For example, caregivers could use media devices at home to play the scenario-based AI early screening program and then use a scoring application with clearly guided rating procedures to evaluate the child’s behavior, thereby enabling early detection of ASD. Furthermore, the integration of AI technology capable of identifying children’s responses in real time could also be explored. Another advantage of such a technology-based screening program is that it can provide immediate results and guidance on next steps in a single, streamlined process, ultimately enabling early detection and intervention for young children.

Second, community-based implementation can be considered. For example, if this tool is applied in primary institutions such as public health centers or childcare support centers, it may be possible to build an efficient large-scale screening system to detect ASD risk in young children. By presenting standardized, scenario-based programs to children, consistent screening can be achieved regardless of the examiner, even when different personnel are involved [50]. This approach may serve as an effective alternative in community settings where professional resources are limited. Furthermore, linking the tool with early childhood health services could help establish a system that monitors developmental status and refers children in need of diagnostic evaluation to specialists, thereby laying the foundation for an effective community-based continuum of screening, diagnosis, and intervention.

Third, the use of a video-based program provides high technological accessibility and scalability. Because the scenariobased ASD early screening tool relies on video presentation, it can be easily implemented with any video-capable device. This makes it particularly useful in areas with geographic or economic constraints [51]. Additionally, because results can be obtained immediately, decisions about early intervention can be made quickly, thereby increasing opportunities for timely intervention [52,53]. Furthermore, the tool could be integrated into community child development and mental health events or public campaigns to raise awareness of early ASD screening and facilitate interagency collaboration as a valuable public health resource.

While the scenario-based ASD early screening tool has shown potential as a primary screening tool for early childhood ASD, there are several limitations and considerations to keep in mind when applying it in real-world settings. First, this study was conducted with a sample of approximately 200 participants and was not a multicenter study, which means that the sample size may be too small to draw definitive conclusions about the tool’s utility as a screening tool. As a result, the sample may not represent children from diverse cultural and social backgrounds. Future studies should increase sample diversity to enhance the generalizability of the findings. Second, while the tool demonstrated high sensitivity, its low specificity raises concerns about the possibility of false positives. The development process in early childhood is influenced by a variety of factors, making it crucial to improve the specificity of the tool in order to accurately distinguish ASD from other developmental delays. To achieve this, integrating additional behavioral observations may be necessary as a complementary solution. Third, the scenario-based ASD early screening tool, based on a highly structured environment and video program, may not fully reflect the child’s natural behavior in everyday life. While it was useful for assessing response characteristics, it may have limitations in evaluating behaviors that involve interactive elements. Therefore, when using this scenario-based ASD early screening tool, it is important to consider these characteristics and apply it with a clear purpose.

In conclusion, the scenario-based content screening tool has shown promise as an effective tool for ASD screening due to its high sensitivity. With improvements in specificity and application in real-life environments, the tool has great potential to evolve into a more effective screening tool in the future.