This study was an open-label, randomized, parallel-group trial. The trial was conducted at the Children’s Hospital Affiliated to Zhengzhou University (Zhengzhou, Henan, China) from March to August of 2022. All participants were screened for eligibility before being randomly assigned to the BrainFit intervention or control group.

The participants were recruited from Children’s Hospital Affiliated to Zhengzhou University. The inclusion criteria were: (1) aged 6-12 years; (2) ADHD diagnosed by clinicians based on The Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5) [29]; and (3) an IQ score >80, measured by the validated Chinese version of the Wechsler Intelligence Scale for Children, Fourth Edition. Exclusion criteria were: (1) presence of comorbid severe psychiatric or physical disorders that may affect attention (eg, autism spectrum disorder, learning disabilities, abnormal blood glucose, anemia, hyperthyroidism, and hyperadrenocorticism) and (2) current or recent (within the last 6 months) medication treatment (eg, methylphenidate, atomoxetine, risperidone, modafinil) or (3) behavioral treatment for ADHD.

Prior to randomization, participants were informed that they would either be assigned to the 4-week intervention group or the waitlist control group. The randomization schedule was generated by an independent statistician not involved in the clinical trial (using the PROC PLAN procedure in SAS 9.4) and the randomization schema was reproducible. Participants were not blinded to group assignment due to the nature of the digital intervention.

The intervention was delivered via iPads preinstalled with the BrainFit software.

Two researchers received training on intervention delivery using BrainFit, which included reviewing the intervention protocol and standardized training manual, practicing intervention delivery steps, and understanding how to respond to potential adverse events. The training material also included a tutorial for BrainFit and a written overview of the study protocol.

At the time of diagnosis, participants’ parents were provided psychoeducation on ADHD according to Chinese consensus standards for ADHD treatment [30]. The intervention was facilitated by trained researchers and consisted of 12 sessions (30 minutes each) delivered via iPad technology over 4 weeks. Each participant received three sessions per week (Monday, Wednesday, and Friday after school) with appropriate training levels tailored to their individual conditions. The intervention included six modules, delivered in the same sequence, with adaptive difficulty levels to increase engagement. Participants earned scores based on completing tasks correctly and on time, and games were upgraded or downgraded based on their performance. Participants and their caregivers received study session reminders regularly to reduce missing data, missed sessions, and attrition. Participants in the control group received no intervention during the active study period other than the previously mentioned psychoeducation.

Demographic data were collected from caregivers, including the caregiver report on the child’s social skills. ADHD symptoms and EF were assessed as primary outcomes at baseline and at 4 weeks post randomization by experienced trainers. Demographic data included sex (female or male), age (years), school report card–based academic performance (quartiles), living arrangement (living with parents or living with parents and grandparents), parent-rated sociability (low, medium, or high), and referral source (ie, teacher, parent, physician, or other). ADHD subtype and severity were defined according to the DSM-5. ADHD symptoms were assessed using the Chinese version of the Swanson, Nolan, and Pelham Questionnaire (SNAP-IV) 26-item parent rating scale [31], which has satisfactory measurement properties regarding test-retest reliability and validity. The SNAP-IV measures ADHD inattention symptoms (items 1-9) and symptoms of hyperactivity/impulsivity (items 10-18). In addition, items 19-26 assess symptoms of oppositional defiant disorder (ODD). Each item is ranked on a 4-point scale (0=“not at all,” 2=“just a little,” 3=“quite a bit,” and 4=“very much”). SNAP-IV is one of the most common methods measuring ADHD symptom severity, in which a higher score indicates more severe symptoms. The Chinese version of the Behavior Rating Inventory of Executive Function (BRIEF) parent form was used to assess EF [32]. The BRIEF has 86 items with each item rated on a 3-point scale (1=“never,” 2=“sometimes,” and 3=“often”). The BRIEF yields an overall Global Executive Composite (GEC) score, which is composed of two indices: the Behavioral Regulation Index (BRI) and Metacognition Index (MI). The BRI has three clinical scales (Inhibit, Shift, and Emotional Control) and the MI is comprised of five clinical scales (Initiate, Shift, Memory, Plan/Organize, Monitor, and Organization of Materials). The BRIEF has good psychometric properties, including test-retest reliability (intraclass correlation=0.68-0.89), internal consistency (Cronbach α=0.74-0.96), and validity (Pearson correlation coefficient=0.41-0.64) [33]; higher scores indicate worse EF.

In a review of 21 intervention studies and their effect on EF in ADHD, a medium effect size was noted (standard mean difference [SMD] 0.611) [34]. We hypothesized a similar effect in the current study. Power analyses determined that a sample size of at least 30 participants per arm would be sufficient to achieve power of 80% with a Type I error rate of .05. Assuming a 20% dropout rate, a total sample of 80 participants was needed (40 per group).

Adverse events were assessed at each session. Participants could also report adverse events by calling a study telephone line. All researchers were trained to record details of any adverse events and a participant would be followed until the event resolved. All potential adverse events were reported and reviewed by the principal investigator. One adverse event was reported wherein a legal guardian withdrew a participant from the study after three sessions due to a fever; this event is unlikely to be related to the intervention.

We used IBM SPSS Statistics for Windows version 23.0 and Stata BE version 17 for all data processing and statistical analyses. We present continuous data as mean (SD) and categorical data as numbers and percentages. To assess any potential differences in participant demographic variables between the intervention and control groups, we used the independent-sample t test and Mann-Whitney U test for continuous variables and ordinal variables; χ² tests were used for comparisons of categorical data. For the multivariable regression analysis of the primary outcomes, SNAP-IV and BRIEF scores were regressed onto the independent variables (treated as dummy variables), including age, sex, whether the participant was the first child, ADHD subtype, academic performance, and sociability ratings. These independent variables were selected based on clinical judgment and the literature. Controlling for these baseline characteristics in the regression model can generally increase the precision of the estimator for the average treatment effect [35]. All independent variables were entered into the model simultaneously. To obtain unbiased intervention effect estimates and quantify the uncertainty surrounding the point estimates, we used the Fisher-Pitman permutation test for two independent samples. This permutation test is a powerful alternative to the Wilcoxon-Mann-Whitney rank-sum test [36,37]. This approach has the benefit of making no assumptions about the distributional properties of the joint distributions of variables as well as minimizing the false-positive results due to multiple testing with traditional regression-based hypothesis testing [38,39]. Because the exact calculation of the test can be extremely computationally intensive, a Monte Carlo–based algorithm (with 10,000 repetitions) was used to increase efficiency with minimal loss of accuracy. We used an α level of .05 for determining statistical significance. The effect size (Cohen d) for each outcome measure was calculated with 95% CIs based on the nonparametric bootstrap method (with 5000 repetitions) to quantify the uncertainty surrounding the point estimates [40].

The study protocol was approved by the Medical Ethics Committee of Henan Children’s Hospital (2021-H-H09) before study initiation. The legal guardians of all participants provided written informed consent and children provided assent to participate before baseline assessment. All families were informed that participation was voluntary, that data collected during the study will be kept confidential, and that they can withdraw their consent at any time without negative consequences to their care. Each participant received ¥600 (US $82.80) as compensation for completing all sessions of the intervention. If a participant completed all assessments, the various scale assessments were offered at no cost to the participants. They also receive an explanatory document and consent form from the researchers that included contact information for any inquiries. The personal information of the participants and their guardians was anonymized, and a unique study ID number was assigned to each participant for data entry, management, and analysis. The individual pictured in Figure 1 provided written informed consent to allow their image to be published.

A total of 199 children were assessed for eligibility and 145 were deemed to be eligible. Among the eligible candidates, 90 children were randomized into either the intervention (n=44) or control (n=46) group. Eighty children completed the study and their data were analyzed, as described in the CONSORT (Consolidated Standards for Reporting Trials) diagram in Figure 2.

Table 1 summarizes the demographic characteristics of participants. The participants’ mean age was 8.4 years, including 63 (78.8%) male participants. The participants’ mean IQ was 98.7 (SD 10.8). The academic performance of more than 65% of the participants with ADHD ranked in the bottom half of the class. Nearly 90% of the participants were rated as having poor social skills by their caregivers and nearly 80% were the first child in the family. More than 60% had a nuclear family model (ie, living with parents only) and were referred to the hospital by a family member. Age, sex, IQ, achievement, social skills, ADHD severity, and prevalence of each ADHD subtype were not significantly different between the two groups of participants (all P>.05).

The primary outcome was ADHD symptoms as measured by the SNAP-IV total and subscale scores. Table 2 shows the results of the parents’ rating of SNAP-IV scores at baseline and at the study endpoint in the intervention and control groups. Participants in the intervention group had a significantly larger reduction in SNAP-IV total scores than those in the control group. For subscale scores, the overall reduction was attributable to significantly lower scores on the Inattention, Hyperactivity/Impulsivity, and ODD subscale scores.

In terms of EF as rated by the BRIEF, participants in the intervention group had a significantly larger reduction on the MI and the GEC, with a smaller and nonsignificant reduction in the BRI (Table 2). Except for the Inhibit clinical scale, children in the intervention group showed a larger reduction across all clinical scales as compared to those in the control group, with five clinical scales showing significant decreases: Initiate, Working Memory, Plan/Organize, Organization of Materials, and Monitor. The Initiate clinical scale decreased from a mean of 63.10 to 54.25 for participants in the intervention group, whereas the control group scores decreased from 59.90 to 59.03 (β=–5.80, P=.005).

Figure 3 depicts the improvements in the context of the wide range of baseline scores for all participants. The effect sizes were generally small to moderate for the outcomes that showed a statistically significant difference in improvements between the intervention and control groups.

This study investigated one of the first digital interventions combining both cognitive and physical (exercise) training to address the core symptoms of ADHD and EF deficits. Compared to those in the control group, children randomized to the BrainFit intervention showed statistically significant improvement in reducing symptoms of ADHD and enhancing EF, as rated by their caregivers.

To our knowledge, this is the first study combining cognitive training and exercise therapy, two effective interventions for elementary and middle school–aged children with ADHD. Findings from this study provide evidence that a digital cognitive-physical intervention could be a promising nonpharmacologic treatment for ADHD.

The therapeutic effects of BrainFit are comparable to those reported in previous studies. A recent systematic review and meta-analysis of 14 trials reported that digital interventions significantly improved attention of children with ADHD (SMD 0.28, 95% CI 0.14-0.41). Our study demonstrated an improved inattention score with a comparable effect (SMD 0.986, 95% CI 0.581-01.447). This evidence suggested a comparable or even greater improvement of inattention than previous digital interventions, which could be related to the application of the exercise intervention. However, our intervention showed disadvantages in certain subdomains of ADHD symptoms. Davis et al [41] reported that the digital intervention “EVO” could improve attention, working memory, and inhibition in children with ADHD. However, in this study, the scores on the Inhibit, Shift, and Emotional Control subscales did not differ significantly between the intervention and control groups, which could be related to the design of the intervention or subjectivity of caregiver ratings.

With the development of digital technologies [42], increasing studies have demonstrated their promising role in the management of various diseases, especially mental and behavioral disorders [43]. Compared with traditional pharmacotherapy and behavioral interventions, digital interventions have the advantages of a minimal requirement of human resources and less adverse effects, which could reach more patients worldwide, especially for patients in less-developed countries and regions. With greater accessibility and compliance, digital therapies could provide proper and timely intervention for children with ADHD, reducing the number of children who do not receive any intervention to manage their symptoms. The US FDA approved a game-based digital therapy called “EndeavorRx” for the treatment of ADHD on June 15, 2020, which demonstrated therapeutic effects on attention among children aged 8-12 years with ADHD [44]. Emerging digital therapies such as BrainFit have the potential to be translated to clinical settings, thereby offering clinicians and families more choices for ADHD intervention.

Cognitive training has been shown to reduce deficits in EF in children with ADHD by strengthening neural networks, while physical exercise increases the levels of catecholamines such as dopamine and norepinephrine [17,27,45,46]. Exercise that is cognitively engaging can increase cognitive abilities in children with ADHD by engaging the regions of the brain involved in EF [16,27]. An “exergame” developed by Benzing and Schmidt [25] was tested in 51 children (aged 8-12 years) in a randomized controlled study that spanned 8 weeks. The game employed strength, coordination, endurance, attention, inhibition, and task-shifting. The treatment group showed improved attention, inhibition, task-shifting, general mental health, and motor abilities compared with those of the control group [25].

In another study, 35 adolescents were assigned to a treatment arm of ADHD medication and 6 weeks of exercise consisting of a sports program or to a control group of ADHD medication and education, and the results showed that the addition of exercise had a strong influence on task-shifting and improving the speed of attention processing [47]. The exercise program consisted of approximately 90-minute sessions three times per week in which there was a 10-minute warm-up period; a 60-minute aerobic exercise that engaged EF such as zigzag running, jumping rope individually and in groups, and basketball; followed by a 10-minute cool-down period [47]. It is thus recommended that game content should be included in any intervention program prescribing exercise, and the exercises should be planned and combined [27,46]. Intensity should be emphasized, while perceptual motor exercises should be combined with cognitive training such as motor planning [16,27]. This strategy has been shown to improve social, perceptual, and cognitive functions in children with ADHD [27,46].

Our study showed promising results for a novel digital cognitive-physical intervention targeting symptoms of ADHD and EF deficits. However, several limitations should be considered in interpreting the findings. First, this was a single-center study with a limited sample size; thus, the study was not powered for subscale analyses across measures. Second, the generalizability of our study findings is limited by our lack of control for ADHD-medication use, ADHD subtype, and age distribution in patients with ADHD. Third, this study measured the outcomes immediately after the conclusion of the 4-week intervention; thus, we are unable to assess the durability of the treatment effects. Based on the encouraging findings of this study, a multicenter efficacy trial controlling for medication use and ADHD subtype with a broader age distribution and a longer follow-up period is planned. Fourth, a dose-finding study with different frequencies of sessions may be necessary to optimize the intervention dose and assess treatment heterogeneity. Finally, the lack of blinding and a sham control may have led to expectancy effects from the parents in completing the assessments.

The findings of this randomized controlled trial conducted in China suggest that BrainFit, a novel digital intervention combining cognitive training and exercise, could improve the symptoms and EF in children with ADHD. The intervention has the potential to increase access to ADHD treatment and improve ADHD symptoms in school-aged children.