This pilot study used single-case experimental methods. A multiple-baseline design across participants was applied to evaluate the program for urination (participants T06, T09, and T10), and a single-case design with baseline, intervention, and follow-up phases was used for defecation training (participant T05).

This study was approved by the Public Institutional Review Board designated by the Ministry of Health and Welfare of South Korea (P01-202308-01-056). All procedures were conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all parents or legal guardians prior to participation.

Children were recruited through an online community for caregivers of children with developmental disabilities. Children were eligible if they were aged 3 to 6 years and had either a clinical diagnosis of ASD by a pediatrician or an ongoing diagnostic evaluation for suspected ASD at enrollment. Children were excluded if they were receiving ongoing toilet training outside the study protocol at enrollment. Eligibility additionally required persistent toileting difficulties (eg, elimination outside the toilet, no spontaneous toilet visits, and/or prior attempts with continued problems); no participation in a structured toileting intervention within the 4 weeks prior to enrollment; and the ability to understand simple instructions, with no reported vision or hearing impairments. Formal standardized measures of imitation ability or receptive language were not collected at screening. Withdrawal and discontinuation criteria were defined a priori and applied consistently: participation could be discontinued if (1) unexpected adverse responses or safety concerns arose, (2) repeated refusal or low engagement prevented feasible delivery of the training procedures and/or adequate recording, (3) a new structured toileting intervention was initiated during the study period, or (4) the caregiver withdrew consent. Four children (T05, T06, T09, and T10; n=2 girls) met the prespecified analytic inclusion criterion (≥5 caregiver-reported app-guided training sessions) and were included in the final analyses. Participant flow is shown in Figure 1.

The intervention was delivered via a mobile app compatible with smartphones and tablets and was implemented by caregivers in naturalistic settings, primarily the children’s home bathrooms. To standardize access and screen size across families, caregivers were provided with a 10.4-inch Samsung Galaxy tablet with the app preinstalled. Caregivers were asked not to introduce new toileting interventions during the study and to keep the children’s usual toileting routines and concurrent therapies as consistent as feasible across phases.

Outcomes were grouped into 4 domains to describe elimination performance, toileting routine performance, caregiver-reported functioning, and implementation.

Treatment fidelity was evaluated using a 7-item checklist developed for this study based on a prior Korean intervention study using a social validity measure [31]. Caregivers completed the checklist on 5 occasions per participant during the intervention period, rated adherence on a 5-point scale, and reported deviations or modifications. The items assessed were (1) preparation of the training environment, (2) arrangement of materials or reinforcers, (3) caregiver instruction to the child, (4) correct software use, (5) waiting for the child’s response, (6) delivery of reinforcement for correct responding, and (7) appropriate responding to incorrect behavior (eg, replay without reinforcement). Fidelity was calculated as (acquired score/total possible score) × 100. The treatment fidelity checklist items can be found in Table S1 in Multimedia Appendix 1.

Social validity was assessed after the intervention using a caregiver questionnaire (9 items, 5-point Likert scale). Items assessed perceived effectiveness, acceptability, safety, feasibility, and overall satisfaction with the toileting program. The questionnaire was developed and refined based on a prior Korean intervention study reporting treatment fidelity procedures [31].

No parallel control group was included in this study; instead, experimental control was supported by the staggered introduction of the intervention across participants.

Caregivers received study information and completed informed consent; the device was distributed; caregivers received 1:1 training on app use and home implementation, including operational definitions for caregiver recording and completion of the Google Forms–based logs; and preintervention questionnaires and recording instructions were completed. During the study, the research team reviewed submitted records weekly and followed up with caregivers via telephone when needed to clarify missing or ambiguous entries.

At baseline, no app-guided training was conducted. Children followed usual toileting routines, and caregivers recorded toileting events and step-level items. Baseline duration was staggered across participants (minimum of 5 days).

During the intervention, caregivers implemented app-guided sessions daily (minimum once per day; the individualized frequency was kept stable within each child) and continued recording. The intervention was introduced sequentially across participants; a participant began the intervention when the preceding participant met an a priori decision rule (mean routine performance across 3 consecutive intervention days exceeding the baseline mean by ≥15 percentage points). The intervention phase ended when the mastery criterion (≥80% for 3 consecutive sessions or days) was met or when the discontinue criterion was met (no meaningful improvement over 15 sessions).

Caregivers completed questionnaires and implementation measures at the end of the intervention period, and a follow-up was conducted 3 months after the intervention and consisted of 3 recording days using the same logs; the app was not used during follow-up.

Descriptive analyses other than nonoverlap of all pairs (NAP) were conducted using R (version 4.3.1; R Foundation for Statistical Computing). Outcomes were summarized descriptively using visual analysis and phase comparisons. No confirmatory hypothesis testing was performed. To supplement visual analysis of single-case data, NAP [32] was computed using Python (version 3.13.5; Python Software Foundation) for each participant as a descriptive nonoverlap index for daily toileting routine completion. Follow-up days were not included in these calculations. For K-SIB-R Personal Self-Care W scores, SD unit deviation from age norms was computed descriptively as (W − age norm mean)/(age norm SD) using manual-based norms.

A total of 10 children were enrolled. Four children (T05, T06, T09, and T10; n=2 girls) met the prespecified analytic inclusion criterion (≥5 caregiver-reported app-guided training sessions) and were included in the final analyses. Reasons for noncompletion or exclusion from the final analytic sample are summarized in Table S2 in Multimedia Appendix 1. Across phases, the number of caregiver log entries was as follows: n=10 at baseline and n=32 during the intervention for T05, n=42 at baseline and n=90 during the intervention for T06, n=51 at baseline and n=141 during the intervention for T09, and n=32 at baseline and n=144 during the intervention for T10. Follow-up consisted of 3 recording days, during which caregiver log entries were recorded for T05 (n=6), T06 (n=9), T09 (n=14), and T10 (n=14). Participant characteristics and baseline toileting profiles are shown in Table 1.

Table 2 summarizes the phase-level percentage of elimination events that occurred in the toilet bowl (in-toilet elimination). Because Table 2 reports target-specific elimination outcomes, the denominators in Table 2 differ from the total number of caregiver log entries reported above. Baseline in-toilet elimination success varied across participants (0%‐23/27, 85.2%). During the intervention, all participants showed higher in-toilet elimination success than at baseline, and all participants showed 100% in-toilet elimination based on the available events during the limited 3-day follow-up observation window.

For T06, in-toilet elimination increased from 28.6% (10/35) at baseline to 73.8% (62/84) during the intervention (+45.24 percentage points; Table 2). Follow-up in-toilet elimination success was 100% (8/8) based on the available follow-up events (limited 3-day observation window).

For T10, baseline in-toilet elimination was 85.2% (23/27) and increased to 93.5% (115/123) during the intervention (+8.31 percentage points; Table 2). Follow-up in-toilet elimination success was 100% (12/12) based on the available follow-up events (limited 3-day observation window).

For T09, baseline in-toilet elimination was 82.2% (37/45) and increased to 97.8% (131/134) during the intervention (+15.54 percentage points; Table 2). Follow-up in-toilet elimination success was 100% (13/13) based on the available follow-up events (limited 3-day observation window).

For T05, baseline in-toilet elimination was 0% and increased to 93.8% (30/32) during the intervention (+93.75 percentage points; Table 2). Follow-up in-toilet elimination success was 100% (6/6) based on the available follow-up events (limited 3-day observation window).

It should be noted that percentages were calculated using target-specific elimination events as denominators. Practice episodes, nontarget elimination events, and entries with insufficient information were excluded.

Daily toileting routine step completion (percentage) for all 4 participants is shown in Figure 3. Because the number of recorded log entries varied by day, day-level percentages were interpreted primarily by patterns over time (eg, level and trend). Visual inspection also indicated differences in the immediacy of effect: routine completion for T09 exceeded all baseline values from the first intervention days, and T05 increased from a stable 0% at baseline on the first intervention day, indicating relatively immediate changes in level, whereas T06 and T10 showed more gradual improvements across the intervention phase.

T06 showed a low and stable baseline followed by a clear upward shift and increasing trend after intervention onset, reaching sustained high performance later in the intervention phase and remaining elevated during follow-up.

T10 showed moderate baseline performance with variability; after intervention introduction, routine completion increased overall with intermittent dips, and follow-up values remained within the higher range observed late in the intervention.

T09 showed baseline variability; during the intervention, routine completion continued to fluctuate but with an overall upward pattern and repeated high-performance days later in the intervention. Follow-up values remained higher than baseline.

T05 showed 0% routine completion during baseline and increasing routine completion during the intervention, with high performance later in training and elevated values during follow-up.

Supplementary nonoverlap analyses using day-level 16-step completion percentages indicated substantial phase separation across participants, with NAP values ranging from 0.9580 to 0.9808 (Table 3). This index complemented the visual analysis by quantifying the degree of nonoverlap between the baseline and intervention phases.

Phase-level distributions of toileting method categories (independent toilet use, caregiver prompting, child communication or indication, and accident) can be found in Table S3 in Multimedia Appendix 1. Overall, accidents decreased for T05, T06, and T09 during the intervention and were not observed at follow-up for any participant. Independent toilet use increased prominently for T05 and T06. Child communication or indication was observed primarily for T09 and T10 during intervention and follow-up, whereas caregiver prompting remained common, particularly for T09.

Baseline characteristics, including K-GARS-2 scores, are shown in Table 1. K-SIB-R outcomes are summarized in Table 4.

On the K-SIB-R Personal Self-Care domain, age-equivalent scores increased from before to after the intervention. To contextualize change relative to age expectations, Personal Self-Care W scores increased for all participants (ΔW=+8 to +15), and deviation from age-based normative means decreased (Table 4). Notably, T06 moved from approximately 3 SDs below the age norm at baseline to approximately 1 SD below after the intervention, and T05 moved to within 1 SD of the age norm after the intervention. Despite improvement, some participants remained at 2 or more SDs below age norms after the intervention (eg, T10).

Exploratory changes in BRIEF-P subscale scores and K-PSI-4-SF percentile ranks can be found in Table S2 in Multimedia Appendix 1.

Treatment fidelity scores, calculated from caregiver-completed fidelity checklists (5 administrations per participant), indicated favorable adherence to intervention procedures. Social validity ratings (Table 5) indicated favorable caregiver perceptions of acceptability, safety, and feasibility and overall satisfaction with the program.

In this pilot study, a caregiver-implemented mobile app that combined stepwise animated video modeling with a structured reward system was associated with improvements in toileting behaviors among young children with ASD. Across participants, the proportion of elimination events occurring into the toilet increased during the intervention and showed 100% in-toilet elimination during the limited 3-day follow-up observation window conducted 3 months after the intervention. In parallel, completion of the 16-step toileting routine improved over time, suggesting gains in broader toileting-related self-care behaviors beyond voiding alone (eg, undressing, flushing, and hygiene steps). Caregivers also reported favorable adherence and favorable perceptions of acceptability and safety, supporting feasibility of home implementation for families who engaged with the app.

A key finding of this pilot study relates to the feasibility boundaries of caregiver-mediated digital interventions. We observed a high attrition rate, with 60% (6/10) of enrolled families discontinuing participation or being excluded from analysis for heterogeneous reasons, including child condition or sleep-related difficulties, low tablet or app engagement, inability to progress beyond toilet sitting readiness, caregiver recording or implementation nonadherence, and postenrollment ineligibility (Table S1 in Multimedia Appendix 1). The high attrition rate represents a major limitation of the present study but also provides useful feasibility information for future research. The pattern of discontinuation suggests that successful implementation of this app requires specific child prerequisites (eg, basic imitation skills and tolerance for tablet interaction) and sufficient parental resources to facilitate the initial sessions. This aligns with broader findings in eHealth research, where open trials often reveal high attrition, highlighting the need to tailor inclusion criteria to the families most likely to benefit [33]. Thus, the findings suggest that the intervention may be beneficial for families able to sustain engagement, but broader applicability requires further study with clearer inclusion criteria. Toileting interventions for children with ASD have historically emphasized in-toilet voiding outcomes using intensive schedules and reinforcement-based packages and may vary widely in components and reporting. Prior reviews indicate that, although many approaches can improve voiding outcomes, the evidence base remains dominated by small-n single-case experimental studies and heterogeneous outcome definitions, which limits comparability and confidence in generalizability [4,5]. Video modeling is often considered well suited for ASD because it reduces verbal demands and leverages visual learning, with evidence supporting its use across functional skill targets [13,14]. Toileting-focused studies have further shown that incorporating animation into video modeling can enable depiction of sensitive behaviors while retaining a complete routine sequence [19,20]. Our findings add to this literature by applying an animated stepwise routine in a mobile format designed for repeated use in naturalistic caregiver-managed settings.

A central conceptual clarification warranted by this study concerns the distinction between the digital platform and the behavioral intervention it delivered. The app served as a structured delivery medium for a multicomponent behavioral intervention that incorporated task analysis, behavior chaining, caregiver-mediated prompting, contingent token-based reinforcement, and stepwise animated video modeling. The study findings therefore reflect the preliminary feasibility and potential benefit of this intervention package as implemented through a caregiver-mediated digital platform rather than evidence on any single technological feature. This distinction is consistent with the broader literature on mHealth interventions, which emphasizes that behavioral mechanisms—not digital packaging—drive therapeutic change. Future work should use dismantling designs or component analyses to clarify the relative contributions of video modeling, token reinforcement, task analytic structure, and caregiver mediation to observed outcomes.

A central design decision was to operationalize toileting as a 16-step task analysis aligned with chaining principles. This allowed caregivers to target discrete steps and reinforce incremental progress rather than focusing exclusively on elimination outcomes [24,25]. Importantly, step completion was tracked even when elimination did not occur (eg, the child sat on the toilet but did not void), reflecting how toileting practice occurs in real-world routines. This may be relevant for children with ASD who often experience difficulty with sequencing and adaptive self-care skills [1]. The choice of animation (rather than live action) also addressed privacy and potential discomfort associated with human depictions of toileting-related behaviors while providing a consistent, child-friendly model.

Although the app included interactive elements, we interpret the motivational component primarily as reinforcement support. The token exchange rule (10 digital stickers exchanged for an immediate reward) implemented a clear contingency while allowing for individualized reinforcers (eg, in-app activities and/or caregiver-selected preferred items). This characterization avoids overstating “gamification” and makes the hypothesized mechanism explicit: repeated practice of a structured routine paired with contingent reward.

Beyond caregiver-recorded toileting logs, standardized assessment data provided descriptive contextual information that was generally consistent with the behavioral observations. On the K-SIB-R, toileting age-equivalent scores increased for all participants, and Personal Self-Care W scores moved closer to age-based expectations (Table 4). Norm-referenced deviation (SD units) decreased for each participant. These SD unit values represent distance from age norms rather than an effect size from a controlled comparison and should be interpreted cautiously. However, reporting a standardized measure alongside event-based and routine-based outcomes is consistent with the observation that the pattern of change was not limited to a single measurement approach.

This study has limitations consistent with early-stage digital health and single-case experimental research. First, the analytic sample was small (4 children), and 60% (6/10) of enrolled participants were not included due to discontinuation or insufficient exposure, underscoring the central importance of engagement and feasibility in caregiver-mediated digital interventions. Because baseline characteristics were not systematically available for all participants who were not included in the final analyses, formal comparison between completers and noncompleters was not possible; however, available contextual information and reasons for noncompletion or exclusion are summarized in Table S1 in Multimedia Appendix 1. Second, primary outcomes relied on caregiver-reported logs, which may introduce reporting bias and incomplete capture of events occurring outside caregiver supervision. Although caregivers received structured training and the research team reviewed records weekly with telephone follow-up when needed, independent observation and interobserver agreement data were not collected, limiting the reliability of caregiver-reported outcomes. Third, the follow-up phase consisted of 3 recording days at 3 months after the intervention, providing a limited sample of maintenance. Finally, day-level step completion percentages can be influenced by variation in the number of recorded events per day; therefore, trends over time are more interpretable than isolated daily values. In addition, technology-related factors may have affected feasibility. Children with limited interest in or tolerance for tablet-based activities may have difficulty engaging with the intervention, and caregiver time, digital literacy, and home technology environments may influence implementation. Therefore, app-based toileting interventions may not be appropriate for all children with ASD without additional engagement supports or caregiver assistance. Furthermore, because this was an early-stage single-case pilot feasibility study, a formal group-level sample size calculation was not performed. The enrollment target was determined pragmatically based on the intensive caregiver-mediated home implementation protocol, the multiple-baseline design, and anticipated recruitment constraints in young children with ASD requiring toileting support. These limitations indicate that the present findings should be interpreted as preliminary signals of feasibility and potential benefit rather than definitive evidence of effectiveness.

Future work should prioritize more rigorous evaluation, improved measurement, and strategies to sustain engagement while preserving the pragmatic, home-based nature of delivery. We have initiated an exploratory randomized controlled trial (target N=30) to evaluate effectiveness and implementation outcomes more robustly. This next phase can follow established reporting guidance for eHealth interventions and incorporate objective use metrics (in-app logs), a longer follow-up period, and independent outcome verification where feasible. Additional iterative development may also be warranted to support sustained engagement over longer periods given that low engagement was a reason for discontinuation in some families.

A caregiver-implemented mobile app that integrates stepwise animated video modeling, task analysis–based chaining, and reinforcement support was feasible for families who engaged with the software and was associated with improvements in toileting outcomes and adaptive self-care indicators among young children with ASD. Larger controlled studies are needed to determine effectiveness, characterize engagement and attrition, and establish scalable implementation strategies.