A Preliminary Study on the Efficacy and Safety of a Novel Digital Therapeutic Device: A Randomized Clinical Trial in South Korean Pediatric Patients With Attention-Deficit/Hyperactivity Disorder

Abstract

Objective

Attention-deficit/hyperactivity disorder (ADHD) is one of the most common neurodevelopmental disorders in children. As digital therapeutics (DTx) emerge as adjuncts to traditional ADHD treatment, we evaluated the efficacy and safety of a novel game-based DTx (ADAM-101) in pediatric ADHD. We hypothesized that adding ADAM-101 to pharmacotherapy would improve ADHD symptoms and attention more than pharmacotherapy alone. This study represents one of the first multicenter DTx trials conducted in South Korea.

Methods

This prospective, multicenter, open-label randomized clinical trial enrolled 54 children aged 7–12 years with ADHD. Participants were assigned to an intervention group (pharmacotherapy+ADAM-101) or a control group (pharmacotherapy only). The intervention group used ADAM-101 for 25 minutes per session, five times a week, for 4 weeks. Primary outcomes were the Korean ADHD Rating Scale (K-ARS) and the Advanced Test of Attention (ATA). Clinical Global Impressions (CGI) scales were used as secondary outcome measures. Safety was assessed via vital signs, electrocardiography, and adverse event reporting.

Results

Compared with control, the intervention group showed significantly greater improvement in parent-rated K-ARS (Cohen’s d=1.07) and the auditory ATA Sensitivity Index (d=0.61). No significant differences were observed in visual ATA measures. CGI scales also favored the intervention group (CGI-Improvement d=1.57; CGI-Severity d=1.05). No serious adverse events were reported.

Conclusion

ADAM-101 as an adjunct to pharmacotherapy showed preliminary benefits in clinician- and parent-rated symptom severity and auditory attention, without significant safety concerns. Larger confirmatory trials are warranted.

INTRODUCTION

Attention-deficit/hyperactivity disorder (ADHD) is one of the most commonly diagnosed neurodevelopmental disorders in childhood, characterized by core symptoms such as inattention, hyperactivity, and impulsivity [1]. Beyond these hallmark features, children with ADHD often exhibit emotional dysregulation, irritability, explosive anger, as well as impairments in social cognition and peer relationships [2]. These symptoms can significantly disrupt daily life and academic functioning, imposing substantial psychosocial and economic burdens on families and communities [3].

In recent years, the diagnosis and management of ADHD have gained increasing attention in South Korea. According to national health insurance claims data, the number of children diagnosed with ADHD nearly tripled between 2019 and 2023, showing a marked increase from 420,053 to 1,122,266 total cases, and a more than twofold increase among children aged 5–9 years (from 126,408 to 262,737). During the same period, total healthcare expenditures related to ADHD rose from approximately 40.4 billion KRW to 139 billion KRW (28 million to 97 million USD). Pediatric treatment costs for the 5–9 age group also more than doubled from 11.8 billion KRW to 28.7 billion KRW (8 million to 20 million USD) [4].

Despite rising awareness and healthcare utilization, pharmacological treatment for ADHD in young children remains controversial due to concerns on side effects, medication adherence, and long-term outcomes [5-9]. Behavioral interventions are widely recommended by leading clinical guidelines, including the American Academy of Pediatrics and the National Institute for Health and Care Excellence, especially for preschool-aged children [10,11]. Moreover, the Centers for Disease Control and Prevention emphasizes ADHD as a chronic condition that requires long-term, family-centered care and classifies affected youth as having special health care needs [12].

Considering these treatment challenges and unmet needs, digital therapeutics (DTx) have emerged as promising adjuncts to traditional interventions [13,14]. DTx are software-based, evidence-based therapeutic tools designed to target specific symptoms, commonly through repeated cognitive-behavioral training. Their scalability, safety, and accessibility at home make them particularly attractive for pediatric use and early intervention. A recent meta-analysis of 25 randomized controlled trials (RCTs) involving 1,780 participants found that digital interventions produced significant reductions in overall ADHD symptom severity (standardized mean difference [SMD]=-0.33; 95% confidence interval [CI]=[-0.51, -0.16]) compared with control conditions [14]. Improvements were observed both in inattention (SMD=-0.31; 95% CI=[-0.46, -0.15]) and hyperactivity/impulsivity (SMD=-0.15; 95% CI=[-0.29, -0.02]), suggesting that DTx may exert a modest but meaningful effect in ameliorating key ADHD symptoms.

However, research on DTx for ADHD in East Asia, particularly in South Korea, remains at an early stage, and the existing literature is still limited in both scope and scale. For example, multicenter randomized studies have recently been reported in Japan, particularly for SDT-001, a localized version of EndeavorRx [14,15]. However, recent studies in China and South Korea have largely been single-site feasibility or pilot investigations with single-center cohorts [16-20]. In particular, the lack of published multicenter RCTs in South Korea prior to the present study limits the generalizability of conclusions regarding the efficacy and safety of ADHD-focused DTx, underscoring the need for rigorously designed multicenter RCTs to guide clinical uptake in East Asia.

The current study evaluates the feasibility and the potential clinical utility of ADAM-101, a novel game-based digital therapeutic device developed for children with ADHD. It is designed to enhance cognitive control processes relevant to ADHD—particularly response inhibition and divided attention—through adaptive, feedback-driven training within a gamified Go/No-Go paradigm. We conducted a prospective, multicenter, randomized, parallel-comparison exploratory clinical trial to examine the efficacy and safety of ADAM-101 as an adjunct to pharmacotherapy. The study assessed both behavioral and neurocognitive features using validated measures, and global functioning scales (e.g., Clinical Global Impression [CGI] scales).

Hypotheses

1) We hypothesized that the combination of the digital therapeutic device and pharmacotherapy would result in greater improvement of ADHD symptoms compared to pharmacotherapy alone, as assessed by the change in Korean ADHD Rating Scale (K-ARS) scores from baseline to week 4.

2) We hypothesized that the combination of the digital therapeutic device and pharmacotherapy would result in greater improvement in attention measures compared to pharmacotherapy alone, as assessed by the change in Advanced Test of Attention (ATA) scores from baseline to week 4.

The clinical trial would be considered successful if statistically significant improvements were observed in both assessment tools (K-ARS and ATA) after 4 weeks.

METHODS

Study design and participants

This study was designed as a prospective, multicenter, openlabel, parallel-comparison exploratory randomized clinical trial. Due to the lack of prior literature providing sufficient data to estimate the expected effect size for the outcome measures, and the exploratory nature of the study, no formal sample size calculation was performed. The target enrollment of 54 participants (27 per group) was determined pragmatically, considering available resources, recruitment feasibility, and in consideration of a similar pilot study of a digital therapeutic intervention for ADHD [21]. Notably, a post-hoc power analysis based on the effect size reported in the pilot study (Cohen’s d=0.71) suggested that approximately 52 participants would be sufficient to ensure adequate statistical power. Thus, our target enrollment of 54 participants was broadly consistent with this estimation, further supporting the adequacy of the chosen sample size.

Eligible participants were pediatric patients between the ages of 7 and 12, who had been previously diagnosed with ADHD by a pediatric psychiatrist, and were currently receiving stable pharmacotherapy, defined a priori as no change in medication type or dose for at least 4 weeks prior to enrollment. All participants were outpatients recruited from three major university hospitals, each located in different areas of South Korea: Seoul, Daegu, and Iksan. Recruitment notices were posted within each hospital to inform eligible patients and caregivers about the study. After a thorough elaboration on the purpose of the study, written informed consent was obtained from both the participants and their legal guardians. Participants were excluded if they had a standardized IQ score of 69 or lower, congenital genetic disorders, or a clear history of acquired brain injury such as cerebral palsy. Additional exclusion criteria included comorbid diagnoses of schizophrenia, bipolar disorder, or other childhood-onset psychoses; seizures or other severe medical or neurological conditions that could interfere with study participation (e.g., sensory impairments, glaucoma, traumatic brain injury); and significant suicidal ideation as determined by clinicians. Enrollment was restricted for those who had participated in another clinical trial within 30 days prior to screening, as well as those unable to communicate or follow instructions adequately. Any participant deemed unsuitable by the investigators for reasons of safety, compliance, or scientific validity was excluded from the study. All participants were first-time users of ADAM-101 and had no prior experience with the device.

Concomitant therapies

During the clinical trial, restrictions were imposed on the use of concomitant medications and therapies to ensure internal validity and to isolate the effects of the digital therapeutic. Prohibited concomitant treatments included antipsychotics, anticonvulsants, and central nervous system-acting medications other than the prescribed ADHD pharmacotherapy. However, participants were allowed to continue certain stable therapies deemed not to interfere with the study outcomes. Medications for non-psychiatric conditions (e.g., asthma, allergic rhinitis) were also allowed if stable and medically necessary. All concomitant medications and treatments were recorded at baseline and monitored throughout the study. Investigators were required to document any changes and assess their potential impact on study results.

Study procedures

The study consisted of three main phases: 1) screening and verifying eligibility (including checking for medical histories and assessment of vital signs), subject enrollment and randomization; 2) baseline assessment (within 14 days after or on the same day as screening) and handing out ADAM-101; 3) post-test assessment (4 weeks+up to 2 weeks of window period).

An independent statistician, not otherwise involved in the trial, generated the randomization sequence using block randomization in (Power Analysis and Sample Size Software 2016). After screening and checking for eligibility, participants were randomly assigned to either the intervention group (pharmacotherapy+ A DAM-101) or the control group (pharmacotherapy only). The randomization list was provided to the sponsor’s quality assurance team, who labeled and packaged the investigational devices according to the sequence. At each site, the investigator verified the participant’s randomization number and corresponding device number, and the device manager dispensed the device with the matching number.

The intervention group received instruction on how to use the digital therapeutic device, “ADAM-101.” They were instructed to use the device for approximately 25 minutes per day, five times a week, for four consecutive weeks (total of 20 sessions). Caregivers monitored adherence through usage logs, and participants were encouraged to minimize exposure to other digital games during the study period. In order to enhance adherence, caregivers received daily text messages and were reminded if the child had not completed the 25-minute usage by preset times of the day. The control group continued pharmacotherapy as usual without additional intervention during the main trial period. However, following the completion of all outcome assessments, control group participants were given an opportunity to use ADAM-101 for four consecutive weeks if they wished, to ensure ethical treatment access and participant equity.

To assess compliance with both the digital therapeutic device and pharmacological treatment, participants were instructed to return the device with a completed adherence log at their last visit. Study staff verified the adherence logs according to predefined compliance criteria: number of sessions completed divided by the total number of prescribed sessions. Participants who used the digital therapeutic device or adhered to medication less than 75% of the prescribed time (i.e., fewer than 15 sessions out of total 20 sessions) were classified as noncompliant and excluded from per-protocol analyses (Figure 1). Safety was assessed via vital signs, electrocardiograms (ECGs), and adverse event reporting.

Figure 1.

CONSORT flow diagram of participant recruitment, allocation, follow-up, and analysis. FSIQ, Full-Scale Intelligence Quotient.

Interim analyses and stopping guidelines

No formal interim analyses were planned for this exploratory trial. However, the protocol prespecified criteria for early termination or suspension of the trial, as well as participant withdrawal, as follows:

• Trial suspension or termination could occur if safety concerns arose that made continuation impractical, upon the recommendation of the principal investigator to the Institutional Review Boards (IRBs), or at the request of a clinician due to safety issues with the investigational device.

• Mandatory suspension was required if participants experienced severe adverse events, or device-related reactions that made continuation impossible, or if a clinician judged that trial termination was necessary for other reasons.

• Participant withdrawal could occur due to voluntary withdrawal of consent, receipt of biological or device-based treatments likely to affect study outcomes without investigator approval (including cessation or change of ADHD medication), failure to meet inclusion/exclusion criteria, adverse events preventing continuation, loss to follow-up, or changes in participant condition making continuation unsafe or unethical.

• Handling of withdrawals included recording the reason for the withdrawal, collecting all available data up to withdrawal, and, unless a valid reason was documented, including the participant in safety and efficacy analyses. For participants missing scheduled visits, follow-up was attempted via written notice or phone calls.

This trial was approved by the IRBs of all participating institutions and by the Ministry of Food and Drug Safety (MFDS) of the Republic of Korea, between November 2023 and January 2024. The trial was also retrospectively registered with the Clinical Research Information Service (CRIS), of the Republic of Korea (KCT0010576) following completion of the trial. Written informed consent was obtained from both the participants and their legal guardians before participation in the study. This clinical trial was approved by the following IRBs: 1) Seoul National University Hospital: IRB No. 2312-114-1494; 2) Daegu Catholic University Medical Center: IRB No. MDCR-23-019; and 3) Wonkwang University Hospital: IRB No. 2023-11-074.

Digital therapeutic device: ADAM-101

The investigational device, “ADAM-101 (a.k.a. Guardians DTx^TM) (Figure 2),” is a game-based digital therapeutic device treating pediatric ADHD patients. It delivers gamified cognitive training through a tablet PC or smartphone. It aims to train various cognitive aspects (e.g., vigilance, reaction time control, stimulus discrimination, focused attention, selective attention, sustained attention, divided attention) via the Go/No-Go task paradigm, with the difficulty level progressively being adjusted in real-time.

Figure 2.

Screenshot from ADAM-101: player (center) should collect as many coins on the map while trying to press the corresponding button on the bottom corners of the screen every time a monster appears on screen (above).

The program consists of two main tasks—the navigation task and the Go/No-Go task—which are presented either separately or in combination across different stages. For the navigation task, the player has to control a character running along a path while collecting or avoiding stimuli (coins) that appear on the road. The player must discriminate between the target and the non-target coins (distractors) based on visual cues (e.g., color) which requires continuous perceptual and response control. The difficulty level of the task, such as the character’s speed and the spacing of the coins, changes throughout the stage in real-time according to the player’s performance. The Go/No-Go task requires identifying target stimuli (monsters) that suddenly appear on the screen and quickly pressing the corresponding button on either the lower left or right corner of the screen. Difficulty of the task is modulated in real-time depending on the player’s performance by several factors: the display time (duration each monster remains visible), the inter-stimulus interval, and the number of non-target monsters (distractors) which differ in both color and appearance.

ADAM-101 also incorporates gamified contents such as animated animal-like characters, stage progression, and reward systems with collectible virtual items that were intentionally developed to address one of the key limitations of prior digital or computerized attention-training programs: low user motivation and poor adherence. By providing a playful and rewarding training environment, ADAM-101 transforms repetitive cognitive tasks into enjoyable gameplay, helping children sustain engagement across sessions and promoting motivation along with higher treatment compliance.

Each session includes 4 to 5 stages, lasting approximately 25 minutes in total. Participants in the intervention group were instructed to complete five sessions per week over a 4-week period (total of 20 sessions). Once a participant completed the daily training quota of at least 25 minutes, further access to the game was temporarily disabled for the remainder of that day to prevent overuse and to maintain consistent training dosage. Likewise, after completing all five sessions within a given week, access to the game was automatically restricted until the start of the next week. For this clinical trial, all participants in the intervention group used (Samsung Galaxy Tab A7 Lite), equipped with Android 14 OS and an 8.7-inch screen. This device was selected for consistency across participants and preloaded with ADAM-101 prior to distribution.

Note on intellectual property and ongoing research

“ADAM-101 (a.k.a. Guardians DTx^TM)” is currently patent pending, and a large-scale confirmatory clinical trial is in progress to further validate its efficacy and safety. Due to the ongoing nature of both the intellectual property process and the confirmatory trial, detailed product specifications and certain proprietary contents cannot be disclosed at this time. Upon completion of the confirmatory trial and subsequent commercialization of the product, comprehensive resources—including user manuals, demonstration videos, and promotional materials—will be made publicly available to facilitate broader understanding and clinical application.

Outcome measures

K-ARS

K-ARS is a caregiver-reported scale based on the ARS by DuPaul and colleagues, and is one of the most widely-used measures for rating ADHD symptoms [22,23]. It consists of 18 items, with odd-numbered items evaluating inattention and even-numbered items assessing hyperactivity-impulsivity. Scores are rated on a 4-point Likert scale, with higher scores indicating greater symptom severity [23].

ATA

ATA is a computerized neurocognitive assessment developed by the Seoul National University College of Medicine, which evaluates attention and impulsivity in children. It consists of visual and auditory continuous performance tests, with each test measuring omission errors, commission errors, reaction time and its variance, Sensitivity Index (d’; a signal-detection-based measure of response discriminability derived from hit and false alarm rates), and criterion bias (β). The tests provide raw scores and standardized values for both visual and auditory tests. Specifically, omission errors, commission errors, response time, and response time variability are converted into T-scores (M=50, SD=10) and Z scores, which are then used to calculate an overall ADHD index score (M=100, SD=15). This composite index provides a standardized indicator of attentional performance relative to age-based norms. ATA is standardized in South Korea and widely used in clinical ADHD assessment [24].

CGI-Improvement and CGI-Severity

The CGI scales were developed as standardized instruments to assess global illness severity and treatment improvement, and are commonly used in psychiatric research [25]. CGI-Improvement (CGI-I) evaluates overall symptom improvement on a 7-point scale (1=very much improved, 7=very much worse), while CGI-Severity (CGI-S) assesses illness severity from 1 (normal) to 7 (among the most severely ill patients). These scales were completed by pediatric psychiatrists.

Statistical analyses

Statistical analyses were conducted on both the full analysis set (FAS) and per-protocol set (PPS) population. Primary efficacy analyses were based on the FAS, while supplementary analyses were performed using the PPS (Supplementary Tables 1-8). Group differences were assessed using analysis of covariance (ANCOVA) with baseline values as covariates. Within-group comparisons utilized paired t-tests or Wilcoxon signed-rank tests, depending on normality which was assessed using the Shapiro–Wilk test. As prespecified, the trial was considered successful only if the difference between the two groups was statistically significant for both co-primary endpoints (K-ARS and ATA). Analyses of secondary outcomes were exploratory and interpreted accordingly without multiplicity adjustment. Missing values in primary outcomes were imputed using the Last Observation Carried Forward (LOCF) method. Significance was set at p<0.05 (two-tailed), and effect sizes (Cohen’s d) were reported where applicable. All statistical analyses were conducted using IBM SPSS Statistics (version 26.0) and R (version 4.3.2) with the following packages: car, psych, emmeans, and dplyr.

Protocol amendments

Within the first month after enrollment of the initial participant, the study protocol was amended once. Initially, the primary outcome measure was limited to ATA scores. However, following the feedback from clinicians emphasizing the importance of capturing overall and global improvements in symptoms of ADHD beyond laboratory-based attention assessments, K-ARS was added as a co-primary outcome measure. This amendment was approved by the Ministry of Food and Drug Safety (MFDS), Republic of Korea, and all relevant Institutional Review Boards (IRBs) prior to implementation. It is worth noting that this amendment was implemented early in the recruitment period and was not informed by interim outcome patterns. The researchers did not have access to accumulated outcome data during enrollment, and efficacy data were not available for analysis until after completion of data collection and database lock.

RESULTS

Participant characteristics

Between February and November of 2024, a total of 56 individuals were screened at three participating sites (Table 1). Two participants failed to pass the screening phase due to not meeting eligibility criteria, leaving 54 participants enrolled (27 per group) in the FAS. A total of five participants were excluded from the PPS: one for low adherence of under 75%, two for violations of eligibility criteria, one for not completing the primary outcome measure (K-ARS), and one for withdrawal of consent. The demographic and clinical characteristics of participants were comparable between groups at baseline, with no significant differences in age, gender, or IQ scores (Table 2; for detailed demographic information see Supplementary Tables 9-11). Regarding stimulant use, 10 of the 54 participants were not receiving stimulant medication (4 in the intervention group and 6 in the control group). The proportion of stimulant use, however, did not differ significantly between the groups (Fisher’s exact test, two-sided p=0.73).

Table 1.

Enrollment result by institutions

Institutions	Screened	Screen failure	Enrolled (FAS)	Drop-out	Completed	PPS
SNUH	18	0	18	2	16	15
DCMC	18	0	18	0	18	18
WKUH	20	2	18	2	16	16
Total	56	2	54	4	50	49

SNUH, Seoul National University Hospital; DCMC, Daegu Catholic University Medical Center; WKUH, Won Kwang University Hospital; FAS, full analysis set; PPS, per-protocol set.

Table 2.

Demographic and baseline characteristics of participants (N=54)

Characteristics	Intervention (N=27)	Control (N=27)	p
Age			0.95
Mean±SD	9.81±1.69	9.78±2.61
Median	9	10
IQR	2	4
IQ			0.46
Mean±SD	94.74±14.31	91.89±13.80
Median	92	89
IQR	14	19
Sex			0.25
Male (%)	25 (93)	21 (78)
Female (%)	2 (7)	6 (22)
Medication			0.73
Stimulant (%)	23 (85)	21 (78)
Non-stimulant (%)	4 (15)	6 (22)

For age and IQ, two-sample t-tests was used; for sex and medication, Fisher’s exact test were used.

Primary outcome measures

K-ARS

After 4 weeks of intervention, the K-ARS total score significantly decreased in the intervention group (M=-5.33, SD=6.73) while the score increased in the control group (M=+1.52, SD=6.10). Between-group comparison of change scores revealed a statistically significant difference in favor of the intervention group (p<0.05, d=1.07). Subscales also showed significant improvements in both inattentive symptoms (p<0.05, d=0.98) and hyperactivity/impulsivity symptoms (p<0.05, d=1.02) in the intervention group (Table 3).

Table 3.

Changes in K-ARS scores from baseline to week 4 (full analysis set)

Outcome	Intervention (N=27)			Control (N=27)			p	Cohen’s d (95% CI)
	Baseline Mean±SD	Week 4 Mean±SD	Mean change (SD)	Baseline Mean±SD	Week 4 Mean±SD	Mean change (SD)
Total	27.59±10.53	22.26±10.79	-5.33 (6.73)	22.59±8.61	24.11±8.80	+1.52 (6.10)	0.001	1.07 (0.49, 1.63)
Inattention	15.48±5.36	12.33±5.43	-3.15 (4.14)	12.78±5.26	13.59±4.77	+0.82 (3.95)	0.005	0.98 (0.41, 1.54)
Impulsivity	12.11±5.91	9.56±5.89	-2.56 (3.25)	9.52±5.05	10.52±5.22	+1.00 (3.71)	0.002	1.02 (0.45, 1.59)

Between-group comparisons were performed using analysis of covariance adjusted for baseline. K-ARS, Korean ADHD Rating Scale; CI, confidence interval.

ATA

The auditory Sensitivity Index (d’), the accuracy in distinguishing an auditory target from non-target stimuli, demonstrated a significant group×time interaction effect. The intervention group showed an improvement (mean change=+0.58), whereas the control group showed a slight decline (mean change=-0.04), with the difference reaching statistical significance (p=0.05, d=0.61) (Table 4). Other auditory ATA scores (commission errors: p=0.00; Sensitivity Index: p=0.02; ADHD index: p=0.04) showed within-group improvements in the intervention group but did not reach statistical significance in between-group comparisons (Supplementary Tables 2 and 3).

Table 4.

Changes in ATA auditory sensitivity index (full analysis set)

	Intervention (N=27)	Control (N=27)	p	Cohen’s d (95% CI)
Mean change (SD)	+0.58 (1.09)	-0.04 (0.95)	0.048	0.61 (0.06, 1.15)

Between-group comparisons were performed using analysis of covariance adjusted for baseline. ATA, Advanced Test of Attention; CI, confidence interval.

Secondary outcome measures

CGI-I

The intervention group showed significantly lower CGI-I scores post-treatment (median=2, IQR=1), indicating greater perceived improvement than the control group (median=3, IQR=1, p<0.001, d=1.57, 95% CI=[0.94, 2.21]).

CGI-S

CGI-S scores decreased in the intervention group (mean change=-0.52, SD=0.58), compared to minimal change in the control group (mean change=-0.00, SD=0.39), with a statistically significant group difference (p<0.001, d=1.05, 95% CI=[0.48, 1.62]).

Safety

No serious adverse events were reported in either group. Vital signs and ECGs remained within normal ranges across all timepoints (Supplementary Tables 12-15). Adverse events were assessed via clinician interviews who had previously established rapport with participating caregivers. These symptoms were not systematically probed; however, none were spontaneously reported during clinician interviews or follow-up.

DISCUSSION

The current study explored the feasibility of a digital therapeutic device as an effective and safe adjunct to pharmacological treatment in children with ADHD. Over a 4-week intervention period, participants in the intervention group showed improvements in parent/clinician-rated ADHD symptoms and some of the attention measures. These results align with previous studies reporting that gamified cognitive training improves neural efficiency and task-related attentional control in pediatric populations [13,16,26]. From a safety perspective, no adverse events or device-related issues were reported in either group, reinforcing the clinical acceptability of the intervention. Adherence was generally high, with 26 out of 27 participants (96%) in the intervention group completing more than 75% of recommended sessions, suggesting feasibility and high compliance.

However, caution is warranted in interpreting results, as some scores of ATA did not reach statistical significance using ANCOVA in comparing the group differences. Specifically, no statistically significant differences between the groups were observed in the visual ATA measures, despite the visual nature of ADAM-101. In contrast, a significant effect between the groups was detected in auditory ATA indices, particularly the Sensitivity Index (d’), which reflects the accuracy in distinguishing an auditory target from non-target stimuli.

This outcome may reflect differences in task characteristics and measurement sensitivity. Compared with visual attention tests that rely on stimulus discrimination based on visual cues, the auditory attention task requires participants to rely solely on auditory input, which can be inherently more demanding and potentially more sensitive to change, particularly in the pediatric ADHD population. In contrast, the visual ATA indices may have been less responsive in this sample, as baseline performance was already within the normal range. More specifically, using the ADHD index of ATA that has an established cut-off score of 123 (mean+1.5 SD), 25 participants in the visual domain and 31 participants in the auditory domain were classified as within the normal range at baseline (Figure 3). This limited headroom may have further constrained the ability to detect post-intervention improvement, especially in the visual domain.

Figure 3.

Boxplots of the attention-deficit/hyperactivity disorder indices (A: visual; B: auditory).

Another possible explanation for the non-significant visual ATA findings is the limited transfer between the trained cognitive processes and the outcome measures. Although ADAM-101 utilizes visually presented stimuli, the training paradigm primarily targets dynamic visuomotor coordination, response inhibition, and divided attention under adaptive conditions. In contrast, the visual ATA test evaluates sustained attention under a fixed and relatively simplified stimulus presentation paradigm. Therefore, improvements induced by ADAM-101 may not directly translate to the specific visual ATA indices, particularly within a short intervention period. The current intervention schedule (25 minutes per session, five sessions per week for 4 weeks) was selected based on prior DTx trials on pediatric ADHD samples [13,16] and also to maximize acceptability among caregivers, given the relative novelty of game-based DTx. However, this dosage and duration, together with the modest sample size typical of an exploratory trial, may have reduced the power to detect broader or durable domain-specific changes. Thus, future trials should evaluate dose–response relationships by varying session frequency and/or extending treatment duration, and may include longer follow-up and larger samples to assess durability and generalization.

Interestingly, the index that demonstrated a significant difference between the two groups (e.g., auditory Sensitivity Index) showed a declining trend in the control group over the 4-week period despite ongoing pharmacotherapy (Table 4). This pattern may reflect natural symptom variability of ADHD patients, which is known to exhibit considerable intra-individual variability across days and contexts. Factors such as sleep quality, emotional state, environmental stimulation, and daily routines may influence day-to-day fluctuations in arousal, motivation, and test-taking conditions, as well as attentional performance and behavioral ratings. Therefore, short-term fluctuations may partially contribute to the observed outcome patterns, particularly within a relatively brief intervention period of 4 weeks. Moreover, when children demonstrate relatively high levels of performance at baseline, subsequent assessments may be less sensitive to detect additional gains. In contrast, the intervention group demonstrated modest improvement, which may suggest that ADAM-101 helped stabilize or modestly enhance attentional performance rather than produce large detectable gains under ceiling-limited conditions.

Another limitation relates to potential reporting biases from caregivers and clinicians. Because this study used an open-label design without a sham digital control, parents who observed their children actively engaging with the intervention may have had heightened expectations regarding symptom improvement. Such expectancy effects could influence caregiver-reported outcomes such as the K-ARS. Similarly, clinicians who were aware of treatment allocation might have been unintentionally influenced when completing global rating scales such as the CGI. On the other hand, since ATA shares conceptual and structural features with Go/No-Go–type paradigms, improvements in ATA may reflect near-transfer or task familiarity. To mitigate these concerns, we utilized both a neuropsychological measure (ATA) that is less susceptible to placebo or expectancy effects [27,28], and parent/clinician-rated scales (KARS and CGI), which may be influenced by reporting bias, as primary outcome measures. Nonetheless, the observed improvements may be interpreted as exploratory signals rather than definitive evidence of efficacy.

The multicenter design also introduced heterogeneity in assessment implementation. Although the study followed a shared protocol, the administration procedures and researcher training/qualification for psychological and neuropsychological assessments could not be fully standardized across sites, which may have increased measurement variability and reduced sensitivity to detect group differences. In addition, gaming exposure prior to intervention (e.g., average daily gaming time and internet/game usage habits) was not systematically assessed at the screening stage. Therefore, potential differences between the groups in prior gaming familiarity—which could influence engagement, learning effects, and responsiveness to a game-based intervention—could not be examined.

Thus, future studies should systematically assess prior gaming exposure and digital media use, as familiarity with game-based environments may influence engagement, learning curves, and responsiveness to digital therapeutic interventions. Standardized operating procedures, centralized researcher training, and prospective collection of gaming exposure data prior to intervention would also improve the robustness of future findings. Longer intervention durations and extended follow-up periods would also allow evaluation of dose–response relationships and the durability of treatment effects. Finally, future studies may, where possible, incorporate blinded assessment and an active or sham digital control condition to better isolate the specific effect of the intervention. Such control conditions would also allow clearer separation of the therapeutic effects of the intervention from nonspecific expectancy or engagement effects.

Despite these limitations and the exploratory nature of this study, this multicenter RCT provides one of the first real-world clinical datasets evaluating a game-based digital therapeutic device as an adjunct to pharmacotherapy in South Korea. The study demonstrated high adherence, favorable safety, and practical feasibility within routine outpatient settings, highlighting the translational potential of DTx beyond tightly controlled laboratory environments. By incorporating both parent/clinician-rated scales (e.g., K-ARS, CGI) and standardized neurocognitive measures (e.g., ATA), this study enabled a multidimensional evaluation of treatment effects from complementary clinical perspectives. These findings suggest that game-based DTx may represent a feasible and scalable adjunctive treatment option for pediatric ADHD within routine outpatient settings. However, given the exploratory nature of this study and the modest sample size, larger confirmatory randomized trials with more rigorous control conditions are required to validate these findings and establish the clinical effectiveness of ADAM-101. Taken together, these findings underscore the feasibility, acceptability, and safety of DTx as adjunctive interventions for ADHD and support their further investigation within real-world pediatric clinical settings.

Supplementary Materials

The Supplement is available with this article at https://doi.org/10.30773/pi.2025.0396.

Supplementary Table 1.Change in K-ARS (Full Analysis Set; N=54)

Supplementary Table 2.Change in ATA Visual test (Full Analysis Set; N=54)

Supplementary Table 3.Change in ATA Auditory test (Full Analysis Set; N=54)

Supplementary Table 4.Change in K-ARS (Per-Protocol Set; N=49)

Supplementary Table 5.Change in ATA Visual test (Per-Protocol Set; N=49)

Supplementary Table 6.Change in ATA Auditory test (Per-Protocol Set; N=49)

Supplementary Table 7.Change in CGI-I scale

Supplementary Table 8.Change in CGI-S scale

Supplementary Table 9.Demographic and Baseline Characteristics (Full analysis set; N = 54)

Supplementary Table 10.Demographic and Baseline Characteristics (Per-Protocol Set; N = 49)

Supplementary Table 11.Prevalence of Comorbid Medical and Surgical Histories

Supplementary Table 12.Reports of adverse events

Supplementary Table 13.Vital Signs at Baseline and Week 4 (N = 54)

Supplementary Table 14.Physical Examination Results at Baseline and Week 4

Supplementary Table 15.Electrocardiogram (ECG) Results at Baseline and Week 4