This study was a parallel-group, open-label, single-center, randomized controlled trial to test the effectiveness of a mobile app–assisted parent training program for reducing behavior problems in children with ASD aged 36-84 months. It followed the CONSORT (Consolidated Standards of Reporting Trials) guidelines. A parent training program based on ABA [17,18] was delivered via mobile devices. For inclusion criteria, children of participants were recruited from the Department of Psychiatry at Seoul National University Hospital and underwent the Autism Diagnostic Observation Schedule-2 (ADOS-2) and Korean Childhood Autism Rating Scale-2 (K-CARS-2), provided demographic information, and completed before the evaluation. Parenting stress levels of the parents and caregivers were also measured. The eligible participants were randomly assigned to either the intervention group or the control group at a 1:1 ratio for 12 weeks. The intervention group received a mobile app parent training program for 12 weeks, while the control group continued with their usual treatment but received no intervention. Both groups were assessed before and after the 12-week period (Figure 1).

Participants with eligible children were recruited from November 9, 2022, to December 8, 2023, at the Department of Psychiatry at Seoul National University Hospital. The inclusion criteria for the children were as follows: (1) being aged between 36 and 84 months of age; (2) meeting the criteria for “autism” or “autism spectrum” classification according to the ADOS-2; and (3) having as a primary caregiver an adult aged at least 19 years. The following were the exclusion criteria for this study: (1) the child was currently taking psychiatric medication for behavioral regulation or similar purposes; (2) the child had a history of congenital or acquired brain damage, such as cerebral palsy; (3) the child had a history of seizure disorders or other neurological conditions; (4) the child had an uncorrected sensory impairment (eg, vision or hearing deficits); (5) the child’s primary caregiver did not possess a smartphone or lacked access to smartphone apps; and (6) the primary caregiver either did not give their consent to join the study or opted to withdraw during the course of the study. All participants provided written informed consent.

The children were divided into low- and high-severity autism groups according to ADOS-2 criteria and randomized into intervention and control groups for each severity group. Randomization was performed using a stratified block randomization method (with a block size of 4) with 1:1 assignment to either the intervention or control group. The randomization list was created using the PROC PLAN method (SAS Institute Inc) in SAS (version 9.4; SAS Institute Inc) software. The randomization process was managed and operated by an independent third party, the Medical Statistics Team of Asan Medical Center.

As this was an exploratory clinical trial, a formal calculation for sample size was not necessary; however, we followed the method described by Julious [28] to calculate the sample size for the pilot trial. The recommended sample size for such pilot studies is 12 persons per group. Considering the possibility of lower compliance with the use of digital therapeutics compared with taking conventional medicines, this study was conducted with a total of 60 participants, 30 in the intervention group and 30 in the control group.

The mobile app–assisted parent training program aims to improve the behavior of children with ASD by identifying the causes of problem behaviors through a self-report questionnaire and providing digital content about it. Previous studies have shown that mobile app or telehealth-enabled ABA interventions can be effective in managing and reducing problem behaviors and increasing parent engagement in children with autism [29-31]. The program was developed in the form of a mobile app to help parents apply ABA-based therapy to their children in their daily lives to reduce their problem behaviors [32-34]. The mobile app–based training program lasted 10-15 minutes per session and was conducted 2-5 days a week for 12 weeks. It covers topics such as an introduction to ABA, identifying the underlying reasons for problematic behavior, creating a behavioral support plan, implementing behavior intervention, practicing in different situations, individualized learning opportunities, and crisis management (Table 1). To further validate our approach and differentiate it from previous studies, we included a system in the mobile app that allows parents to record and monitor their child’s behavior and progress. This real-time data collection allows for continuous monitoring and adjustment of the intervention to ensure responsiveness and effectiveness. Program content was developed with the consultation of 2 ABA experts and 1 child psychiatry specialist.

The mobile app–assisted parent training program was available for download from a mobile app store, and participants accessed this study’s intervention using a unique, preannounced username and password. Parents in the intervention group completed the mobile app parent training program at home over a 12-week period, during which they continued to receive their usual nondrug treatment in addition to the training program. The mobile app included recording problem behaviors and viewing the frequency of child problem behaviors, education, and homework assignments to help caregivers generalize skills in a real-world setting (Figure 2). All sessions included instruction on the parent training program, video content with case examples to help parents understand the program, and quizzes and homework to check for understanding of the session. Participants could access the app at any time, and all programs delivered were relearnable.

The control group continued to receive their usual nonpharmaceutical treatments, such as behavioral therapy, speech therapy, and play therapy, for 12 weeks, without receiving the mobile app–assisted parent training program intervention.

Research data were stored in a secure laboratory and personal information was stored in a separate file to ensure that personal identification is not possible through the research data. Enrollment logs were used to maintain personal identification information separately; the case report form did not include personal information except for initials and case numbers. This helped to protect participants’ privacy.

All outcomes were analyzed in participants who completed the program period and assessments at baseline and after the intervention. The analysis included participants who did not violate the protocol (per-protocol analysis set). Owing to the nonnormal distribution, nonparametric methods were used in this study. All baseline demographic variables and evaluation outcomes at baseline and after the intervention were summarized by randomized groups using the median (IQR) for continuous data or count (%) for categorical data. We estimated the median values and IQRs for each group and time point using the pre- and postintervention evaluation scores for the intervention and control groups. Before testing the effectiveness of the program, the Wilcoxon signed rank test was performed to determine whether there were differences in demographic and baseline variables between the groups. The Wilcoxon signed rank test was used to compare the effects of all outcomes. To evaluate the intervention effects between the 2 groups, the Kruskal-Wallis test was performed by judging the rejection range based on a significance level of 0.05. To assess differences in session completion rates within the intervention group, we estimated medians and IQRs and used Wilcoxon signed rank tests. To examine the correlation of completion rates with evaluation outcomes, Pearson correlation coefficient was used. Data analysis was performed using the R software (version 4.1.0; R Foundation) and Python (version 3.9.12; Python Software Foundation).

Written informed consent was obtained from all participants. This study was approved by the Seoul National University Hospital Institutional Review Board (H-2205-158-1329).

All data were stored in a locked laboratory, and participants’ personal information was stored in a separate file from the research data to prevent identification through the research data. Enrollment logs were maintained to keep personal information separate and to ensure that personal information was not exposed. Only initials and case numbers were used on case report forms to further protect participant identity.

The tests administered as part of this study were provided free of charge. Compensation was not contingent on the participant completing this study, and pretest results were provided even if the participant did not complete this study. The posttest was only administered if the participant completed this study. No other monetary compensation, including transportation, was offered.

Between November 9, 2022, and December 8, 2022, a total of 64 participants were enrolled. Further, 56 participants who met our inclusion criteria were randomly assigned into the intervention and control groups; however, after randomization, 8 participants in the intervention group and 3 in the control group declined to participate, resulting in 20 participants receiving the intervention and 25 participants in the control group. In the control group, 3 participants were excluded from the analysis; these included children who started a new treatment during the intervention period. Finally, a total of 42 participants were included in the data analysis (Figure 3).

Demographic and baseline variables were compared between the intervention and control groups before the intervention. Of the 42 participants, 20 (48%) were in the intervention group and 22 (52%) in the control group. There were no significant differences in sex, age, severity, or ethnicity between the 2 groups. Demographic variables, including the K-CARS-2 and ADOS-2, did not differ significantly between the groups. The current treatment also did not differ significantly between the intervention (n=17, 85%) and control (n=18, 81.8%) groups. Neuropsychological testing showed no significant differences between the 2 groups. The demographic and baseline variables of both groups are summarized in Table 2. Complete information is presented in Table S1 in Multimedia Appendix 1. Table S2 in Multimedia Appendix 1 provides demographic and baseline information for the full sample using an intention-to-treat (ITT) analysis approach. The results of this ITT analysis revealed a significant difference between the intervention group (14.5, 95% CI 12.5 to 16) and the control group (12, 95% CI 10 to 15) on social effect, a subscale of the ADOS, compared with the results obtained from the per-protocol analysis (P=.045). Except for this item, no significant differences were found for the remaining items. Multimedia Appendix 1 provides more details.

The results for the K-CBCL scores are shown in Table 3. According to the K-CBCL, lower scores indicate a decrease in overall problem behaviors. The K-CBCL total score for the intervention group decreased from baseline to after the intervention (median difference –0.5, 95% CI –4.5 to 3). In comparison to the control group, the intervention group’s median withdrawn score difference reduced to –2 (95% CI –8 to 4), median difference of attention decreased by –1 (95% CI –6 to 4), and median difference in pervasive developmental disorder exhibited a –1 (95% CI –5 to 2) decrease. The full table is shown in Table S3 in Multimedia Appendix 1.

There were no serious intervention-related adverse events that led to treatment discontinuation in either group (Table 5).

This study tested the feasibility of a mobile app–assisted parent training program by comparing pre- and postintervention outcomes over a 12-week intervention to evaluate the effectiveness of the program in reducing behavioral problems in children with autism. Specifically, this study demonstrated the potential for a mobile app–based intervention to reduce behavioral problems in children with ASD. These findings can help in the planning of early interventions and high accessibility of mobile app–assisted parent training programs for children with ASD who have behavioral issues. Problem behaviors in children with ASD cause additional challenges for parents and create uncertainty about how to manage these behaviors. Given these challenges, parents of children with ASD can handle behavioral issues through parent training [46].

Regarding this study’s primary outcome, we found no clinically important differences between the intervention and control groups. A possible explanation for these results is that both groups maintained their existing treatments, such as ABA therapy, sensory integration therapy, and language therapy, which may have provided an indirect pathway for improving problem behaviors in children with ASD. However, the results can still be considered clinically significant. The pre- and posttreatment K-VABS-2 scores for the intervention group showed clinically significant effects for both communication and socialization. Daily living skills also showed clinically significant effects when comparing the intervention and control groups (Table 2). These scores represent the impact of overall problem behaviors, adaptation, and social performance in children, with similar results to previous studies [47,48]. Considering the exploratory nature of this study, we used the K-CBCL as the primary outcome measure, which provides subscale scores across various measurements. However, the actual significant effects were observed in core areas wherein individuals with ASD face difficulties, such as socialization, as was evident in the K-VABS-2, a widely used measure of adaptive behaviors in children with developmental disorders. These findings suggest the potential for conducting a confirmatory study using more focused outcome measurements.

Results on parenting stress showed that, within the intervention group, caregivers who used the training program the most reported higher levels of parenting stress (Table 3 and Figure S2 in Multimedia Appendix 1). This may imply that the more parenting stress a caregiver has, the more extensively they use the program.

Analyzing the results based on session completion rates, a more pronounced difference was observed between K-CBCL and K-VABS-2. While the scores measured by K-CBCL did not exhibit significant variations based on different completion rates, a notable difference was found in the K-VABS-2 scores. One of the key advantages of digital interventions is their relative freedom from constraints such as time, space, and cost. Therefore, a higher session completion rate would have been expected. However, considering that the session completion rate in this study was not as high as anticipated, our data suggest the need for complementary strategies to enhance user engagement. In doing so, if parental participation in this training program increases, our study’s findings indicate the potential for further improvements in the K-VABS-2 scores.

Ongoing research suggests that a significant number of parents are interested in mobile app–based, evidence-based parent training programs, which do not rely on therapists because the content is standardized and can be easily accessed by parents anytime and anywhere [49]. Using these mobile app technologies can help bring key elements of evidence-based interventions to populations that may not otherwise be able to receive them.

This study had several limitations. First, it relied on parental opinions to assess children’s behavior. Although the CGI-S and CGI-I were assessed by a physician, as the physician’s evaluation relied on parental reports regarding the child, it was also influenced by the parents’ opinions. Second, we observed a nonsignificant clinical effect of the treatment group on the primary outcome, which we attribute to indirect effects from other treatments the child was already receiving owing to this study’s nature. Third, our study had low program participation rates. For studies based on mobile app interventions, there are challenges in promoting participant engagement in the program.

This study demonstrated the feasibility of using mobile devices for evidence-based parent training to reduce problem behaviors in children with ASD. In addition to the high prevalence of ASD and the high cost of treatment, there is a significant shortage of people to provide treatment, which is a barrier to treatment delivery. The accessibility and flexibility of mobile devices may make them a viable alternative as a parent education tool in providing early intervention for problem behaviors in children with ASD.