Medina Et Al., (2021)

Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

Original Paper

Electrophysiological Brain Changes Associated With Cognitive


Improvement in a Pediatric Attention Deficit Hyperactivity Disorder
Digital Artificial Intelligence-Driven Intervention: Randomized
Controlled Trial

Rafael Medina1, MSc; Jaime Bouhaben1, MSc; Ignacio de Ramón1,2,3, MSc; Pablo Cuesta2,4, PhD; Luis Antón-Toro2,5,
MSc; Javier Pacios2,5, PhD; Javier Quintero6, MD, PhD; Josep Antoni Ramos-Quiroga7,8,9, MD; Fernando Maestú2,5,
PhD
1
Sincrolab Ltd, Madrid, Spain
2
Laboratory of Computational and Cognitive Neuroscience, Centre for Biomedical Technology, Polytechnic University of Madrid, Pozuelo de Alarcón,
Spain
3
Faculty of Health, Camilo Jose Cela University, Villafranca del Castillo, Spain
4
Radiology Rehabilitation and Physiotherapy, Complutense University of Madrid, Madrid, Spain
5
Department of Experimental Psychology, Faculty of Psychology, Complutense University of Madrid, Madrid, Spain
6
Department of Psychiatry, University Hospital Infanta Leonor, Madrid, Spain
7
Department of Psychiatry, Hospital Universitari Vall d’Hebron, Barcelona, Spain
8
Group of Psychiatry, Mental Health and Addictions, Vall d’Hebron Research Institute, Barcelona, Spain
9
Biomedical Network Research Centre on Mental Health, Barcelona, Spain

Corresponding Author:
Ignacio de Ramón, MSc
Sincrolab Ltd
Prensa 7
Madrid, 28033
Spain
Phone: 34 630 364 425
Email: nacho@sincrolab.es

Abstract
Background: Cognitive stimulation therapy appears to show promising results in the rehabilitation of impaired cognitive
processes in attention deficit hyperactivity disorder.
Objective: Encouraged by this evidence and the ever-increasing use of technology and artificial intelligence for therapeutic
purposes, we examined whether cognitive stimulation therapy implemented on a mobile device and controlled by an artificial
intelligence engine can be effective in the neurocognitive rehabilitation of these patients.
Methods: In this randomized study, 29 child participants (25 males) underwent training with a smart, digital, cognitive stimulation
program (KAD_SCL_01) or with 3 commercial video games for 12 weeks, 3 days a week, 15 minutes a day. Participants completed
a neuropsychological assessment and a preintervention and postintervention magnetoencephalography study in a resting state
with their eyes closed. In addition, information on clinical symptoms was collected from the child´s legal guardians.
Results: In line with our main hypothesis, we found evidence that smart, digital, cognitive treatment results in improvements
in inhibitory control performance. Improvements were also found in visuospatial working memory performance and in the
cognitive flexibility, working memory, and behavior and general executive functioning behavioral clinical indexes in this group
of participants. Finally, the improvements found in inhibitory control were related to increases in alpha-band power in all
participants in the posterior regions, including 2 default mode network regions of the interest: the bilateral precuneus and the
bilateral posterior cingulate cortex. However, only the participants who underwent cognitive stimulation intervention
(KAD_SCL_01) showed a significant increase in this relationship.
Conclusions: The results seem to indicate that smart, digital treatment can be effective in the inhibitory control and visuospatial
working memory rehabilitation in patients with attention deficit hyperactivity disorder. Furthermore, the relation of the inhibitory

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 1


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

control with alpha-band power changes could mean that these changes are a product of plasticity mechanisms or changes in the
neuromodulatory dynamics.
Trial Registration: ISRCTN Registry ISRCTN71041318; https://www.isrctn.com/ISRCTN71041318

(J Med Internet Res 2021;23(11):e25466) doi: 10.2196/25466

KEYWORDS
ADHD; cognitive stimulation; magnetoencephalography; artificial intelligence; Conners continuous performance test;
KAD_SCL_01; AI; cognitive impairment; attention deficit hyperactivity disorder; pediatrics; children; rehabilitation

unknown. With the aim of providing evidence, we examined


Introduction whether a CRB digital training regimen would be effective in
Inhibitory control deficit is one of the core impairments in an ADHD child population after 12 weeks using the continuous
attention deficit hyperactivity disorder (ADHD) [1,2]. This performance test (CPT) inhibitory control measure as the main
deficit is directly related to the levels of impulsiveness present outcome. We hypothesized that after the intervention, the
in the symptoms of ADHD [1,3-5] and produces difficulties in inhibitory control, as a core symptom of ADHD, would show
the everyday activities of those afflicted [6] while adversely a better performance and that this would be related to changes
affecting academic performance [7]. According to the literature in the alpha band in the posterior regions and the DMN
reviewed, other impairments can be found in ADHD including according to magnetoencephalography (MEG). We also tested
the performance of cognitive processes, such as working whether treatment-produced changes in secondary outcomes
memory [8], sustained attention [9,10], alternating attention would be related to ADHD and, finally, whether it could
[11], and planning [12,13]. decrease the clinical symptoms associated with ADHD and
change those behaviors related to executive functioning.
In ADHD, the indices of inhibition, task switching, and
emotional control appear to be related to relative power values Methods
of the alpha frequency band (7-13 Hz) in midline brain regions
measured at resting state [14,15] and with performance in The study was approved by the local ethics committee of the
attentional tasks [16]. These patients consistently present a San Carlos Hospital (Madrid, Spain). All legal representatives
decrease in the alpha band in the central and posterior regions of the participants gave their written informed consent to
[17-24], as well as an increase in the theta frequency band (3-7 participate in the study. This clinical trial is registered in the
Hz) and the theta: beta ratio [17-21,25-28]. The decrease of the ISRCTN registry (ISRCTN71041318).
alpha band in regions that engage the default mode network
(DMN; active network at resting state which includes the
Participants
caudate nucleus, medial prefrontal cortex, posterior cingulate A total of 41 children diagnosed with combined-type ADHD
cortex, hippocampus, inferior parietal lobe, cerebellum, and (ADHD-C) were recruited (34 males). Contact with participants’
precuneus) could modulate impairments in the functional legal guardians was made through health facilities, schools, and
connectivity of this network [29-33]. These impairments in the associations in the community of Madrid. Research staff first
DMN also seem to be related to inhibitory control deficits contacted those private and public clinical centers asking for
[34,35]. permission and agreement to recruit. The order in which centers
were contacted was at random. The following recruitment
Cognitive stimulation therapy appears to be effective in patients actions were performed: emailing study information, phone
with ADHD [36]. The progressive increase of the workload in calls, and teleconferences and webinars with legal guardians
cognitive stimulation tasks is one of the main treatment summarizing study information. Participants’ legal guardians
dynamics of this type of therapy [37], and there are many who agreed to participate authorized communications with
examples of its effectiveness in ADHD and other disorders research staff. Eligibility criteria were checked by phone and
[38-42]. Its effectiveness seems to stem from the fact that these email with legal guardians prior to visit 1. Before any other
increases in the workload in cognitive tasks trigger an increase study activity, legal guardians read and signed an informed
in long-distance connections supported by alpha and beta bands, consent. There were no artificial intelligence (AI) requirements
and a decrease in short-distance connections supported by delta for the eligibility.
and theta bands [43-46].
To be eligible, participants had to meet the following 5 criteria:
Although the increase of the workload in cognitive stimulation (1) aged 8 to 11 years; (2) diagnosis of ADHD-C by an
tasks has shown promise in neurocognitive rehabilitation in authorized professional (chartered psychiatrists at the medical
children with ADHD, a case-based reasoning (CBR) system college); (3) cessation of ADHD medication 3 days before each
[47] that allows the adaptive workload to increase for each visit day, as, according to the technical specification of the drug
patient has never been used. The CRB system has been methylphenidate (Concerta), it has a half-life of 3.5 hours (90%
successful in various clinical areas [48-50], but its efficacy in is excreted in urine and 1 to 35 in feces as a metabolite at 48-96
a digital treatment for rehabilitation of neurocognitive alterations hours); (4) maintenance of the same level of medication during
and its relation to electrophysiological dynamics and its efficacy the at-home intervention period; and (5) compliance with the
in the rehabilitation of clinical alterations in ADHD remain intervention protocol.

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 2


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

ADHD diagnosis was performed by accredited expert intervention (dextroamphetamine, levoamphetamine,


professionals following the Diagnostic and Statistical Manual lisdexamphetamine, methylphenidate, atomoxetine) also made
of Mental Disorders, Fifth Edition (DSM-5) criteria. These participants ineligible for the study. The compliance was
diagnostic criteria were the same across all participants. The checked at the beginning and the end of the participant’s
average time from ADHD diagnosis confirmation to study participation through the child's legal guardians.
enrollment was 2.58 (SD 1.21) years.
From the initial pool of 41 volunteers, 40 were randomly
Participants meeting any of the following 5 exclusion criteria allocated into 1 of the 2 trial conditions (experimental or
were dropped from the trial: (1) the initiation or abandonment control). Of these, 28% (n=11) dropped out during the
of behavioral therapies or psychoactive drugs during the at-home intervention period (control=8, experimental=3). One participant
intervention period; (2) motor difficulties which made the use did not the meet inclusion criterion of stopping ADHD
of the mobile device (tablet or smartphone) impossible; (3) use medication prior to treatment. The Consolidated Standards of
of psychoactive drugs (such as benzodiazepines) which could Reporting Trials (CONSORT) 2010 flow diagram is presented
have acted as a confounding factor, presence or suspicion of in Figure 1.
substance abuse for the past 6 months; (4) presence of blindness
From the final sample of 29 participants, 20 were taking
or uncorrected visual acuity difficulties; and (5) any additional
pharmacological interventions (experimental=9, control=11),
psychological diagnosis.
7 participants were taking nonpharmacological interventions
The inclusion criterion at the level of input data for the AI was such as psychological interventions (experimental= 4,
a diagnosis of ADHD-C by an authorized professional in order control=3), and 4 were taking both (experimental=2, control=2).
to register the patient on the platform. The pharmacological interventions were based on
methylphenidate (Concerta; n=7), methylphenidate (Equasym;
The use of other psychoactive drugs different from those
n=7), methylphenidate hydrochloride (Medikinet; n=2),
approved by the Spanish Agency of Medicines and Medical
lisdexamphetamine (Elvanse; n=2), and methylphenidate
Devices or European Medicines Agency for pediatric ADHD
hydrochloride (Rubifen; n=2).

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 3


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

Figure 1. Consolidated Standards of Reporting Trials flow diagram.

Digital Intervention
Experimental Design
This was single-center, parallel, single-blind, randomized Experimental Condition
controlled trial that examined a pediatric population (8-11 years) KAD_SCL_01 games are designed to work on different
diagnosed with ADHD of combined presentation. It was cognitive processes with an increase of the cognitive load
conceptualized as a proof-of-concept study intended to assess following evidence that the brain’s reconfiguration networks
the preliminary efficacy of a digital, videogame-like, cognitive seem to be fixed by this type of training routine [43-46,53]. The
stimulation therapy, as well as its safety and engagement. 14 games which compose the KAD_SCL_01 cognitive
Proof-of-concept trials are useful in the framework of novel intervention are described in Multimedia Appendix 1. The game
drugs and devices, so knowledge regarding their administration level is adapted based on a case-based reasoning algorithm.
(eg, dosing, user instructions) may be acquired in small samples This algorithm and the human-AI interaction are described in
in order to develop larger clinical trials [51,52]. Multimedia Appendix 2.
Control Condition
Participants received a sham intervention composed of 3
videogames which were not specifically designed to improve

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 4


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

cognitive performance [54]. The specifications are described 60 minutes. The resting-state MEG was also recorded during
in Multimedia Appendix 3. The sham intervention tasks are visits 1 and 2. The order in which participants received both
accessible through Kongregate open-access platform was counterbalanced.
(Kongragate Inc).
The intervention allocation was created by a nonblinded
Main Outcome Measure and Magnetoencephalography Sincrolab researcher (RM) and performed with a simple
randomization function, with a ratio of 1:1 and an allocation
Main Outcome Measure probability of 0.50. Intervention allocation was performed once
The main outcome measure of this study was the change in the eligibility criteria were met, according to the 2010
score found in the commission score from Conners CPT CONSORT statement [57].
(CPT-III) between both groups’ differences (pre- and
The intervention was scheduled for 12 weeks, with 3 sessions
postintervention) [55]. Commissions in CPT-III as a measure
(15-20 minutes each) per week in both groups. The whole
of inhibitory control was chosen as main outcome measure due
intervention period was telematically monitored. Both the
to its use as an efficacy intervention measure in several previous
KAD_SCL_01 and sham control platforms allowed for a daily
studies about the methylphenidate effect in ADHD [56].
checking of performed sessions for a nonblinded Sincrolab
Magnetoencephalography researcher (RM). The number of weekly intervention sessions
Neurophysiological data were acquired using a whole-head performed by the participants was monitored to ensure
Elekta-Neuromag MEG system with 306 channels (Elekta AB) compliance with the 12-week intervention protocol. Safety and
at the Center for Biomedical Technology (Madrid, Spain). MEG adequacy (the number of games played and the consecutive
data were collected at a sampling frequency of 1000 Hz and extreme punctuations of 0 or 100 in the performance, which
online band-pass filtered between 0.1 Hz and 330 Hz. could reflect an issue in the calibration of the AI outputs) were
also assessed. Legal guardians were contacted by study staff in
Head shape was defined relative to 3 anatomical locations order to report any adverse event.
(nasion and bilateral preauricular points) using a 3D digitizer
(Fastrak), and head motion was tracked through 4 head-position Right after the at-home intervention period was over,
indicator coils attached to the scalp. Eye movements were participants who achieved at least 80% completion of
monitored by a vertical electrooculogram assembly composed intervention sessions (28 alongside the 12 prescribed weeks)
of a pair of bipolar electrodes. were appointed for postintervention assessment with same
characteristics as the preintervention one. After the
Other Cognitive Outcome and Clinical Outcome postintervention assessment, the participants who were allocated
Measures in the control arm were offered training with the KAD_SCL_01
The secondary cognitive outcome, aimed at measuring other for 12 weeks.
several aspects of cognitive processing, and clinical Statistical Analyses
questionnaires on ADHD behavioral symptoms and executive
functioning in daily activities are included in Multimedia Data analysis in this proof-of-concept randomized trial followed
Appendix 4. a per-protocol approach [58]. A per-protocol population was
defined as any participant who had been randomly allocated to
Safety and Compliance 1 of the 2 conditions (experimental or control), complied with
Intervention safety was assessed through adverse events. at least an 80% completion of scheduled sessions (28 of 36),
Potential adverse events were monitored and recorded during and had received the postintervention assessment.
the intervention period. Intervention dropouts were also recorded Statistical Analyses of Cognitive Outcome Measures
in order to assess compliance with intervention protocol.
Descriptive statistics of average, distribution shape, and scatter
Study Procedure were calculated. Standardized statistics of asymmetry and
The study procedure occurred in 4 stages: recruitment and kurtosis were used to assess the normality assumptions of each
screening, preintervention assessment (visit 1), at-home distribution. These standardized statistics are calculated by
intervention, and postintervention assessment (visit 2). dividing the statistic between its SE.
Recruitment and screening were carried out as described in the Next, cognitive outcome measures which did not deviate from
Participants section. The details of the Al are described in normality were adjusted to mixed-effects models. Each model
Multimedia Appendix 5. was adjusted with a random intercept and fixed slope (due to
Preintervention and postintervention assessments were the number of repeated measures). An unstructured covariance
performed at the Center for Biomedical Technology, at the matrix (Sigma) was estimated for the random effect factor.
Technical University of Madrid. Assessments were carried out Robust restricted maximum likelihood was chosen as the
by a blinded Sincrolab researcher (JB) who only knew the estimation method of preference due to its robustness with small
number associated with the participant. Assessments including samples and its capability to estimate an unbiased parameter
neuropsychological batteries and MEG recordings were matrix in the presence of missing values. A stepwise method
administered in the same order as reported here. Questionnaires was used for age as a demographic covariable in the main
for the clinical outcome measures were filled out by the legal outcome’s mixed model as a method applied to explicative
guardians. The cognitive assessment lasted for approximately models.

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 5


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

As the commission score from CPT-III was set as main outcome to a criterion of spatial and frequency adjacency. Each cluster
measure, only 1 comparison was performed (1 dependent consisted of several adjacent nodes, which systematically
variable). Therefore, no correction for multiplicity was applied. showed a significant partial correlation (with age as the
Regarding the rest of the cognitive outcome measures, every P covariate) in at least 4 consecutive frequency steps (a 1-Hz
value under significance α value of .05 was taken as statistically interval) between their corresponding power ratio values and
significant due to the exploratory nature of this pilot study. Still, CPT ratio (Spearman correlation coefficient P value <.05).
P values were corrected for multiple comparisons under a Importantly, all nodes within a cluster needed to show the same
Bonferroni correction within a statistical family. The outcome sign of the correlation coefficient for the cluster to be considered
measures from the different cognitive processes (ie, visuospatial a functional unit. Only clusters involving at least 1% of the
working memory) were treated as independent statistical families nodes (ie, a minimum of 12 nodes) in each frequency step were
for Bonferroni adjustments. considered. Cluster-mass statistics were assessed through the
sum of the Spearman ρ values across all nodes and significant
Effect sizes greater than 0.4 (considered as the minimum
frequency steps.
practical effect size [59] in the experimental condition but not
in the control condition) were highlighted. Likewise, for the To control for multiple comparisons, the entire analysis pipeline
main outcome, the predictive positive value (PPV) was was then repeated 5000 times, with the correspondence between
estimated, as the small sample size could have led to power ratio estimates and CPT ratio being shuffled across
overestimation of the effect size. Due to the novelty of this type participants. At each repetition, the maximum statistic of the
of training methodology, a priori effect size and unspecified surrogate clusters (in absolute value) was kept, creating a
prestudy odds (R=0.5) were used in order to estimate the PPV. maximal null distribution that would ensure control of the
familywise error rate at the cluster level. Cluster-mass statistics
Respondent analysis was also performed over the main cognitive
on each cluster in the original data set were compared with the
outcome measure (commission score on CPT-III) in order to
same measure in the randomized data. The network-based
study the proportion of participants per intervention arm who
statistics P value represented the proportion of the permutation
achieved a pre-post difference of at least 0.64 SD, according to
distribution with cluster-mass statistic values greater or equal
other literature [56]. Moreover, with consideration to this a
to the cluster-mass statistic value of the original data.
priori effect and because the estimated sample size could not
be achieved, post hoc power analysis for the mixed model’s Power ratio values were averaged across all nodes and
interaction component was carried out with 200 simulations, frequencies that belonged to the cluster. This average was
and PPV was computed following the procedure in Button et considered to be the representative MEG marker value for that
al [60]. cluster and further participated in subsequent correlation
analyses. Therefore, the statistics presented in the results section
Statistical Analyses of Clinical Outcome Measures was derived from the correlation between the averaged power
Clinical outcome measures were standardized according to ratio value of each significant cluster and the corresponding
Behavior Rating Inventory of Executive Function (BRIEF) and CPT ratio for each participant. As already mentioned,
Evaluación del Trastorno por Deficit de Atención e correlations were first performed within the whole sample. In
Hiperactividad (EDAH) standardized scores (T scores). a second step, correlations between the average power ratio and
Paired-samples t tests were performed over each outcome the CPT commission ratio scores were performed independently
measure and in each intervention group. Respondent analysis for both intervention conditions within the sample (experimental
was also performed on the EDAH outcome measures by and control). Statistical analyses were carried out using
counting the proportion of participants who reached the cutoff MATLAB R2020b (Mathworks Inc).
point of pathology set by the interpretation of the EDAH manual
for each condition. BRIEF and EDAH were treated as Sample Size Justification
independent statistical families for Bonferroni adjustments. A priori sample size was estimated to detect a standardized
mean difference of 0.64 SD in the commission score from the
Magnetoencephalography Signal Preprocessing and CPT-III [56], with a significance level of α=.05 and a power of
Statistical Analyses 0.8 (1-β=.8). The calculation procedure followed the sample
With the intention of facilitating this paper’s interpretation, size estimation for a 2-tailed, 2-samples mean difference with
signal preprocessing analyses are detailed in Multimedia a correction factor for repeated measures [63]. The total sample
Appendix 6. size required was 56, but the actual sample was 29. Nevertheless,
sample sizes of between 10 and 15 participants per condition
Regarding the statistical analyses of the MEG preprocessed
are well-supported in similar literature [64-66].
signal, the aim of this study was the detection of any robust
correlation between power ratio values derived from the clusters The last enrolled participant ended study procedures in February
of nodes localized in certain brain regions and CPT’s 2020. With the COVID-19 crisis and the consequences in Spain
commission ratio (CPT commission postintervention or CPT (since March 2020), the study sponsor and principal investigator
commission preintervention). The goal of this methodology (FM) decided to stop the recruitment procedures due to
included the extraction of any neurophysiological markers whose difficulties and in order to assure protocol compliance in 2020.
dynamic could be associated with the evolution of the Therefore, assuming the exploratory nature of this pilot
inhibition-control performance. Such analysis relied on randomized trial, it was decided that the statistical analysis plan
network-based statistics [61,62]. Clusters were built according be applied to the presented sample.
https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 6
(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

Baseline demographics and other characteristics in each group,


Results as well as the between-group comparison, are shown in Table
Demographic and Baseline Characteristics 1. No significant differences were found between groups.

Table 1. Demographic characteristics in the experimental and control conditions.


Characteristic Experimental group, (%)a Control group, (%)a t or X2 P value
(N=15) (N=14)
Age (years) 9.2 (1.21) 9.71 (1.33) 1.09 .27b
Males 13 (44.8) 12 (41.4) 0.005 .94c
Using medication 9 (31) 11 (37.9) 1.17 .28c
Receiving psychological treatment 4 (13.8) 3 (10,3) 0.11 .74c

a
The characteristic of age is expressed as mean (SD).
b
P values are from a t test (between-participant, 2-tailed).
c
P values are from a chi-squared test (2-tailed).

evolution trend (pre-post training) of each participant and the


Safety and Compliance distribution of each group is shown in Figure 2. No statistically
Three adverse events were reported by legal guardians during significant difference was found between the conditions
the at-home intervention period (Multimedia Appendix 7). (experimental and control) in the preintervention (baseline)
Dropout (n=11) details are shown in Multimedia Appendix 8. measures (t27=1.72; P=.10). Critical ratios for skewness
Main Outcome indicated no deviations in skewness or kurtosis in the normal
distribution.
Descriptive statistics for the main outcome measure in each
condition at each study period are shown in Table 2. The

Table 2. Descriptive statistics for main outcome measure commission score on Conners continuous performance test (CPT-III).
Descriptive statistic Treatment group Control group
Pretreatment Posttreatment Pretreatment Posttreatment
Mean (SD) 53.87 (8.37) 47.80 (8.21) 48.79 (7.53) 49.64 (7.32)
Asymmetry –0.37 –0.17 0.28 –0.09
Kurtosis –0.61 –0.76 –0.33 –1.52

CRa asymmetry –0.32 –0.15 0.23 –0.08

CR kurtosis –0.27 –0.34 –0.14 –0.66

a
CR: critical ratio.

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 7


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

Figure 2. Main efficacy outcome: individual and average change in commission errors from Conners continuous performance test per condition.

Mixed-effects models for main outcome measured with and an improvement of at least 0.64 standardized points. In the
without interaction effects were estimated with the robust control arm, this percentage was just 21.42% (3/14). More
restricted maximum likelihood procedure. The stepwise details about respondent analysis are shown in Multimedia
introduction of the condition-period interaction effect Appendix 11.
significantly improved the model adjustment (X21=4.596; A post hoc power analysis yielded a statistical power of 43%
P=.03). Standardized mean difference (β estimator) for the (1 – β=0.43) for the detection of the condition-period interaction
condition-period interaction effect in the final model (Figure effect and a PPV of 0.81. A priori effect size was used to
2) was statistically different from 0 (β=.86; SE 0.39; t27=2.21; simulate post hoc power rather than observed effect size.
P=.04). The standardized mean differences (β estimators) for
model comparison (baseline model to final model) are shown Magnetoencephalography Outcomes
in Multimedia Appendix 9. Comparison criteria (Akaike A significant cluster (P=.04) was found in the frequency interval
information criterion and Bayesian information criterion) (11.67-13.33 Hz) mainly comprising the posterior regions of
between the models, in addition to model performance statistics the brain (see Figure 3A and Table 3).
(R2 and adjusted R2), are also reported in Multimedia Appendix The power ratio in all frequencies of this interval negatively
9. The graphical diagnosis for the final model with interaction correlated with the CPT ratio across the whole sample
effect is shown in Multimedia Appendix 10. (ρ=–0.562; P=.003). The maximum cluster size was found at
Pre-post standardized mean differences per condition were 12-12.33 Hz (51 nodes). The cluster size oscillated between a
calculated as Hedges g statistic for effect size. A large pre-post minimum of 50 nodes at the beginning of the frequency range
standardized mean difference (g>|0.4|) was found in the and 16 at the end of that frequency range (see Figure 3B).
experimental group (g=–0.62), but not in the control group Furthermore, 12 Hz showed the highest average correlation
(g=0.1). A high PPV (PPV=0.81) was found to be related with coefficient value across all nodes of the cluster ρ= –0.547).
the pre-post standardized mean difference. The correlation between the CPT commission ratio and the
Respondent analysis for the main outcome measure shows that power ratio (11.67-13.33 Hz) in the interval within the cluster
53.33% (8/15) of the experimental participants (KAD_SCL_01 generated in the previous step remained significant for the
intervention) achieved the a priori clinically meaningful effect: experimental group (ρ=–0.783; P=.004; Figure 3C) but not for
the control group (ρ=–0.358; P=.21; Figure 3C).

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 8


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

Figure 3. Brain region whose magnetoencephalography alpha power (11.67-13.33 Hz) was found significantly correlated with CPT commission ratio.
(A) Brain regions within the significant cluster (depicted in blue). (B) Evolution of the cluster size through the different frequency steps (maximum size
at 11.75 Hz). (C) Scatter plot showing the Spearman correlation coefficient between the cluster’s average power ratio and CPT commission ratio and
each subgroup of the sample. CPT: continuous performance test; Freqs: frequency steps.

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 9


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

Table 3. The automated anatomical labeling atlas ROIsa that were partially captured by the significant cluster.b

ROI Portion of ROI occupied, n/N (%)c


Left precuneus 11/28 (39.29)
Right precuneus 8/21 (38.10)
Left cuneus 7/11 63.64)
Right cuneus 7/13 (53.85)
Right superior parietal gyrus 6/18 (33.33)
Left cingulate gyrus, posterior part 3/5 (60.00)
Right superior occipital lobe 3/10 (30.00)
Left superior parietal gyrus 2/16 (12.50)
Right cingulate gyrus, posterior part 1/4 (25.00)
Left calcarine fissure and surrounding cortex 1/20 (5.00)
Left superior occipital lobe 1/11 (9.09)
Right middle occipital lobe 1/17 (5.88)

a
ROI: region of interest.
b
Regions of interest are from the Anatomical Labeling Atlas that are part of the significant cluster where the continuous performance test commission
ratio correlates with power in the alpha band.
c
N is the number of magnetoencephalography sources in our head model that are contained within the ROI volume; n indicates how many sources,
among the corresponding N, are enclosed within the significant cluster; and % is the percentage of each ROI that was captured by that cluster.

P=.02), behavioral composite index (t14=2.62, P=.02), and


Other Cognitive and Clinical Outcomes
general executive composite index (t14= 2.7, P=.01). No
Descriptive analysis for each secondary outcome measure is significant differences in the sham intervention group were
shown in Supplementary Material (Multimedia Appendix 12). found. The experimental arm (KAD_SCL_01 intervention)
No statistically significant differences were found between showed statistically significant pre-post mean differences in all
conditions (experimental and control) in preintervention EDAH measures (hyperactivity score P=.05), inattention score
measurement. (P=.001), behavior disorder score (P=.001), and global score
The mixed-effects model analysis was performed the for main (P=.001). The control arm also showed statistically significant
outcome measure. Only, the backward span score (from the pre-post mean differences in inattention score (P=.001),
Corsi block-tapping test) as a dependent variable (X21=4.64; hyperactivity + inattention, and global score (P=.002), but not
in hyperactivity or behavior disorder score. Respondent analysis,
P=.03) was significant. The standardized mean difference for
descriptive analysis, t statistics, P values, CIs, and respondent
the condition–moment interaction effect (β estimator) in the
percentage per score in the EDAH scale are detailed in
final model was statistically different from 0 (β =–.84; SE 0.38,
Multimedia Appendix 16.
t27=–2.24; P=.03). Multimedia Appendix 13 shows the graphical
representation of the average pre- and postintervention No statistically significant differences were found between
standardized scores for each intervention group (experimental conditions (experimental and control) in the preintervention
and control) in this outcome measure. Standardized mean measurement either in the BRIEF or the EDAH outcome
differences and CIs for cognitive secondary outcomes are shown measures.
in Multimedia Appendix 14, as classified by the cognitive
process each measures (inhibitory control, cognitive flexibility, Discussion
working memory, short-term memory, attention, speed
processing, and verbal fluency). Empirical evidence points suggests that cognitive stimulation
based on progressive workload increments leads to
Effect sizes of g>0.4 in the experimental group but not the improvements in cognitive performance [38-42,67], along with
control group were found in 19 cognitive secondary outcome beneficial regulation of cortical activity patterns [43-46,53].
measures, plus the main outcome. In contrast, only 1 cognitive
secondary outcome measure showed a greater effect size (g>0.4) The results in our study indicate that cognitive intervention
in the control group compared to the experimental one. See triggers significant improvements in inhibitory control in child
Multimedia Appendix 15 for the complete analysis. and adolescent patients with ADHD as measured by Conners
CPT-III. Moreover, this improvement in inhibitory control
The results in the parent version of the BRIEF questionnaire seems to be similar to that found in pharmacological studies on
showed statistically significant pre-post mean differences, the effectiveness of methylphenidate [68]. Meanwhile, the effect
favoring the KAD_SCL_01 intervention participants in shifting size of our study (g=0.62) is consistent with that found in the
score (t14=2.32; P=.03), working memory score (t14= 2.43, meta-analysis by Losier et al [56] on the effectiveness of the

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 10


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

drugs used in ADHD. Therefore the digital treatment proposed generates interference in the task-oriented network, producing
in the present study could be a therapeutic option complementary impairments in the performance of patients with ADHD [79].
to the pharmacological route. From this perspective, this digital cognitive stimulation
intervention, based on progressive workload increases governed
Despite there being no significant differences between the
by CBR algorithms, might be effective for the treatment of
groups in the measure of previous treatment (P=.09), the
ADHD.
possible differences between both groups could be producing
a type I error or false-positive result. However, as observed in One of the secondary objectives of the study was to measure
Figure 2, within the range of 1 to –1 SD, 7 of the 10 patients the effectiveness of the intervention on a set of cognitive
who received the KAD_SCL_01 treatment show an processes that are usually part of the ADHD deficits. The results
improvement in their performance (70%). On the contrary, 6 indicate that cognitive intervention triggers improvements in
of 9 participants belonging to the control group, in the same visuospatial working memory total score. Moreover, practical
range, show worse scores in the postintervention measure. minimum effect size (g> 0.4) was observed in visuospatial
working memory span and in visuospatial working memory
Although several studies have reported that digital cognitive
total score, while numeric working memory total score or
exercises do not show effects superior to those found in other
numeric working memory span seemed to be not be affected
commercial video games not intended for therapeutic uses
by the treatment. These results are similar to those found by
[37,69,70], these findings, like those reported by Davis et al
another digital study, which based its training on empirical
[41] and Kollins et al [42], seem to indicate that adaptive digital
principles [41]. In our study, no significant differences were
training, built on a proven empirical basis, could be effective
found in other measures. However, these results might be due
for the treatment of ADHD.
to a type II or false-negative error, since small sample sizes
The relationship between alpha-band power and performance frequently generate this type of error [80]. Indeed, we found a
in tasks involving attentional and inhibitory control processes practical minimum effect size in 12 of the 42 secondary outcome
has been published in recent publications [14-16,71,72]. In order indices, which is a possible indicator of treatment efficacy. In
to clarify the relationship between the changes in inhibitory contrast, the control group showed equivalent effects in just 1of
control and the possible changes in alpha band power—given the 42 secondary measures (see Multimedia Appendix 15 for
its association with performance in inhibitory control tasks detailed information).
[14,15]—we completed a MEG registry of the participants of
Finally, in this study, the parent version of the BRIEF
both groups. The results seem to indicate that there is a direct
questionnaire was used to obtain a measure of executive
relationship (ρ=–0.56; P=.003) between the improvement in
functions in everyday life. The results indicate that cognitive
inhibitory control and the alpha-band power in the posterior
training triggers significant changes in flexibility, working
brain areas. These changes in the power of brain oscillations
memory, and the composite indices of both behavior and
appear to be associated with brain plasticity processes [73], as
executive functions. These changes appear to be similar to the
well as changes in the dynamics of neuromodulators such as
effects of methylphenidate-based pharmacological treatments
dopamine [74] that are affected in these patients. This
[81] and treatments administered by clinical professionals for
relationship remained significant when the experimental group
executive functions [82].
(ρ= –0.78; P=.004) was analyzed separately, but this was not
the case with the control group (ρ= –0.35; P=.21). This suggests In the ADHD rating scale, in the overall rating, 60% (9/15) of
that the improvements produced in the experimental group are the participants who underwent cognitive training exceeded the
strongly associated with the previously mentioned plasticity cutoff point (<30), compared to 21% (3/14) who worked with
and neuromodulation phenomena. commercial video games, which seems to indicate that this type
of cognitive training may have positive effects on the behavioral
Although this was intended as a power study, we believe our
impact of the disorder.
results are relevant to the functional connectivity literature due
to the participation of the precuneus and posterior cingulate In conclusion, this study reports the preliminary results of a
cortex in the main cluster examined of this paper. These 2 digital cognitive stimulation intervention in a population with
regions of interest conform to the posterior part of the DMN. ADHD. The results suggest that such treatment is effective at
Consequently, the alpha-power increment linked with the CPT improving inhibitory control and visuospatial working memory
decrement may be associated, under our interpretation, with an in patients with ADHD. Moreover, this improvement was
improvement in the functional integrity of the DMN. The observed in the executive measures of daily life and was
decrease in alpha-band power in regions of the DMN could be associated with a reduction of symptoms.
mediating the impairments present in ADHD in the functional
The main limitation of the study relates to the small size of the
connectivity of this network [75-77] and its neurocognitive
sample (N=29) compared to the a priori calculated sample size
correlate [34,35,78]. This effect seems to be due to frequency
(N=56). Consequently, the statistical power was lower than the
band having special importance in the communication between
one desired a priori. Therefore, these results must be interpreted
the regions of this network [33]. In this regard, the data from
as the first evidence of a digital treatment using CBR algorithms,
Sonuga-Barke and Castellanos [79] seems to indicate that the
and more extensive studies are needed to confirm the findings
decrease in connectivity between the regions of the DMN
of this proof-of-concept study.

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 11


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

Acknowledgments
This study was funded by Sincrolab and partly funded by the Centre for the Development of Industrial Technology of the Spanish
Ministry of Economy, Industry, and Competitiveness.

Authors' Contributions
Author contribution: RM, JB, IR, JQ, and FM designed research; JB, LAT, and PC conducted the research; RM, JB, LAT, and
PC analyzed data; and RM, JB, IR, LAT, JQ, JARQ, and FM wrote the paper.

Conflicts of Interest
Sincrolab provided financial support in the form of salaries used for partial salary support for the authors. PC received punctual
financial support for carrying out the magnetoencephalography analysis. Sincrolab participated in the study design, data analysis,
decision to publish, and preparation of the manuscript. IR is the cofounder of Sincrolab. JQ and JARQ are members of the
Scientific Board of Sincrolab. JQ is also a shareholder of Instituto Neuroconductual de Madrid Ltd and a speaker on the advisory
board for Takeda & Jansen. He also receives investigation funding from the Carlos III Health Institute. JARQ was on the speakers’
bureau and/or acted as a consultant for Janssen-Cilag, Novartis, Shire, Takeda, Bial, Shionogi, Sincrolab, Novartis, Bristol Myers
Squibb, Medice, Rubió, Uriach, and Raffo in the last 3 years. He also received travel awards (air tickets and hotel) for taking part
in psychiatric meetings from Janssen-Cilag, Rubió, Shire, Takeda, Shionogi, Bial, and Medice. The Department of Psychiatry
chaired by JARQ received unrestricted educational and research support from the following companies in the last 3 years:
Janssen-Cilag, Shire, Oryzon, Roche, Psious, and Rubió. RM and JB are employees of Sincrolab. The other authors have no
conflicts of interest to declare.

Multimedia Appendix 1
Dynamics, integrated cognitive processes, and hierarchical or multilevel structure of Sincrolab’s games.
[PDF File (Adobe PDF File), 1011 KB-Multimedia Appendix 1]

Multimedia Appendix 2
Case-based reasoning explanation and human-artificial intelligence interaction.
[PDF File (Adobe PDF File), 310 KB-Multimedia Appendix 2]

Multimedia Appendix 3
Description of sham control intervention.
[PDF File (Adobe PDF File), 84 KB-Multimedia Appendix 3]

Multimedia Appendix 4
Cognitive and clinical secondary outcome measures.
[PDF File (Adobe PDF File), 137 KB-Multimedia Appendix 4]

Multimedia Appendix 5
Study procedure diagram.
[PDF File (Adobe PDF File), 71 KB-Multimedia Appendix 5]

Multimedia Appendix 6
Magnetoencephalography signal preprocessing procedure.
[PDF File (Adobe PDF File), 101 KB-Multimedia Appendix 6]

Multimedia Appendix 7
Adverse events.
[PDF File (Adobe PDF File), 29 KB-Multimedia Appendix 7]

Multimedia Appendix 8
Dropouts' details.
[PDF File (Adobe PDF File), 28 KB-Multimedia Appendix 8]

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 12


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

Multimedia Appendix 9
Comparison between performance and standardized mean differences (beta estimators) for each main outcome model.
[PDF File (Adobe PDF File), 36 KB-Multimedia Appendix 9]

Multimedia Appendix 10
Graphical diagnosis for main outcome's final model with interaction effect.
[PDF File (Adobe PDF File), 27 KB-Multimedia Appendix 10]

Multimedia Appendix 11
Comparison of individual performance and clinical effect.
[PDF File (Adobe PDF File), 10 KB-Multimedia Appendix 11]

Multimedia Appendix 12
Descriptive statistics for secondary outcome measures.
[PDF File (Adobe PDF File), 61 KB-Multimedia Appendix 12]

Multimedia Appendix 13
Efficacy outcome in visuospatial working memory: mean change in backward span score from Corsi block-tapping test. Significant
differences were found for condition × moment interaction effect in this cognitive measure (β=–.84; SE=0.38;
<italic>t</italic><sub>27</sub>=–2.24; <italic>P</italic>=.03).
[PDF File (Adobe PDF File), 44 KB-Multimedia Appendix 13]

Multimedia Appendix 14
Standardized mean differences for interaction effects in secondary outcome measures.
[PDF File (Adobe PDF File), 77 KB-Multimedia Appendix 14]

Multimedia Appendix 15
Pre-post standardized mean differences (Hedges g) per treatment group.
[PDF File (Adobe PDF File), 44 KB-Multimedia Appendix 15]

Multimedia Appendix 16
Descriptives, <italic>t</italic> statistics, <italic>P</italic> values, CIs, and respondent percentage per score in the Evaluación
del Trastorno por Deficit de Atención e Hiperactividad scale.
[PDF File (Adobe PDF File), 21 KB-Multimedia Appendix 16]

Multimedia Appendix 17
CONSORT-eHEALTH checklist (V 1.6.1).
[PDF File (Adobe PDF File), 631 KB-Multimedia Appendix 17]

References
1. Willcutt EG, Doyle AE, Nigg JT, Faraone SV, Pennington BF. Validity of the executive function theory of
attention-deficit/hyperactivity disorder: a meta-analytic review. Biol Psychiatry 2005 Jul 01;57(11):1336-1346. [doi:
10.1016/j.biopsych.2005.02.006] [Medline: 15950006]
2. Coghill DR, Seth S, Matthews K. A comprehensive assessment of memory, delay aversion, timing, inhibition, decision
making and variability in attention deficit hyperactivity disorder: advancing beyond the three-pathway models. Psychol.
Med 2013 Oct 31;44(9):1989-2001. [doi: 10.1017/s0033291713002547]
3. Barkley RA. Behavioral inhibition, sustained attention, and executive functions: constructing a unifying theory of ADHD.
Psychol Bull 1997 Jan;121(1):65-94. [Medline: 9000892]
4. Brown TE. Executive functions and attention deficit hyperactivity disorder: implications of two conflicting views. International
Journal of Disability, Development and Education 2007 Jan 22;53(1):35-46 [FREE Full text] [doi:
10.1080/10349120500510024]
5. Barkley RA. Is executive functioning deficient in ADHD? It depends on your definitions and measures. The ADHD Report
2011 Aug;19(4):1-9,16. [doi: 10.1521/adhd.2011.19.4.1]

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 13


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

6. Duff CT, Sulla EM. Measuring executive function in the differential diagnosis of attention-deficit/hyperactivity disorder:
Does it really tell us anything? Appl Neuropsychol Child 2015 Sep 25;4(3):188-196 [FREE Full text] [doi:
10.1080/21622965.2013.848329] [Medline: 25257827]
7. Best JR, Miller PH, Naglieri JA. Relations between executive function and academic achievement from ages 5 to 17 in a
large, representative national sample. Learn Individ Differ 2011 Aug;21(4):327-336 [FREE Full text] [doi:
10.1016/j.lindif.2011.01.007] [Medline: 21845021]
8. Fenollar-Cortés J, Navarro-Soria I, González-Gómez C, García-Sevilla J. Cognitive Profile for Children with ADHD by
Using WISC-IV: Subtype Differences? // Detección de perfiles cognitivos mediante WISC-IV en niños diagnosticados de
TDAH: ¿Existen diferencias entre subtipos? REV PSICODIDACT 2014 Nov 24;20(1):157-176 [FREE Full text] [doi:
10.1387/revpsicodidact.12531]
9. Byrne JM, Bawden HN, DeWolfe NA, Beattie TL. Clinical assessment of psychopharmacological treatment of preschoolers
with ADHD. J Clin Exp Neuropsychol 1998 Oct;20(5):613-627. [doi: 10.1076/jcen.20.5.613.1121] [Medline: 10079039]
10. Zalsman G, Pumeranz O, Peretz G, Ben-Dor DH, Dekel S, Horesh N, et al. Attention patterns in children with attention
deficit disorder with or without hyperactivity. ScientificWorldJournal 2003 Dec 13;3:1093-1107 [FREE Full text] [doi:
10.1100/tsw.2003.94] [Medline: 14625396]
11. Rodriguez-Jimenez R, Avila C, Jimenez-Arriero MA, Ponce G, Monasor R, Jimenez M, et al. Impulsivity and sustained
attention in pathological gamblers: influence of childhood ADHD history. J Gambl Stud 2006 Dec;22(4):451-461. [doi:
10.1007/s10899-006-9028-2] [Medline: 16912931]
12. Seidman LJ, Biederman J, Monuteaux MC, Weber W, Faraone SV. Neuropsychological functioning in nonreferred siblings
of children with attention deficit/hyperactivity disorder. Journal of Abnormal Psychology 2000 May;109(2):252-265 [FREE
Full text] [doi: 10.1037/0021-843x.109.2.252]
13. Álvarez-Arboleda L, Rodríguez-Arocho W, Moreno-Torres M. Evaluación neurocognoscitiva del trastorno por déficit de
atención e hiperactividad. IRD Éditions 2003:83-90 [FREE Full text] [doi: 10.4000/books.irdeditions.25484]
14. Swartwood JN, Swartwood MO, Lubar JF, Timmermann DL. EEG differences in ADHD-combined type during baseline
and cognitive tasks. Pediatric Neurology 2003 Mar;28(3):199-204. [doi: 10.1016/s0887-8994(02)00514-3]
15. Zhang D, Johnstone SJ, Roodenrys S, Luo X, Li H, Wang E, et al. The role of resting-state EEG localized activation and
central nervous system arousal in executive function performance in children with Attention-Deficit/Hyperactivity Disorder.
Clin Neurophysiol 2018 Jun;129(6):1192-1200. [doi: 10.1016/j.clinph.2018.03.009] [Medline: 29653296]
16. Shi T, Li X, Song J, Zhao N, Sun C, Xia W, et al. EEG characteristics and visual cognitive function of children with attention
deficit hyperactivity disorder (ADHD). Brain Dev 2012 Dec;34(10):806-811. [doi: 10.1016/j.braindev.2012.02.013]
[Medline: 22459254]
17. Clarke AR, Barry RJ, Bond D, McCarthy R, Selikowitz M. Effects of stimulant medications on the EEG of children with
attention-deficit/hyperactivity disorder. Psychopharmacology (Berl) 2002 Nov;164(3):277-284. [doi:
10.1007/s00213-002-1205-0] [Medline: 12424551]
18. Clarke A, Barry R, McCarthy R, Selikowitz M, Brown C. EEG evidence for a new conceptualisation of attention deficit
hyperactivity disorder. Clinical neurophysiology 2002;113(7):1036-1044. [doi: 10.1016/s1388-2457(02)00115-3] [Medline:
12088697]
19. Clarke AR, Barry RJ, McCarthy R, Selikowitz M, Croft RJ. EEG differences between good and poor responders to
methylphenidate in boys with the inattentive type of attention-deficit/hyperactivity disorder. Clinical Neurophysiology
2002 Aug;113(8):1191-1198. [doi: 10.1016/s1388-2457(02)00147-5]
20. Clarke AR, Barry RJ, McCarthy R, Selikowitz M, Brown CR, Croft RJ. Effects of stimulant medications on the EEG of
children with Attention-Deficit/Hyperactivity Disorder Predominantly Inattentive type. International Journal of
Psychophysiology 2003 Feb;47(2):129-137. [doi: 10.1016/s0167-8760(02)00119-8]
21. Bresnahan SM, Barry RJ, Clarke AR, Johnstone SJ. Quantitative EEG analysis in dexamphetamine-responsive adults with
attention-deficit/hyperactivity disorder. Psychiatry Res 2006 Mar 28;141(2):151-159. [doi: 10.1016/j.psychres.2005.09.002]
[Medline: 16343642]
22. Barry RJ, Clarke AR, Hajos M, McCarthy R, Selikowitz M, Bruggemann JM. Acute atomoxetine effects on the EEG of
children with attention-deficit/hyperactivity disorder. Neuropharmacology 2009 Dec;57(7-8):702-707. [doi:
10.1016/j.neuropharm.2009.08.003] [Medline: 19698723]
23. Choi H, Park S, Kim KK, Lee K, Rhyu H. Acute effects of aerobic stretching, health and happiness improving movement
exercise on cortical activity of children. J Exerc Rehabil 2016 Aug;12(4):320-327 [FREE Full text] [doi:
10.12965/jer.1632602.301] [Medline: 27656629]
24. Huang C, Huang C, Hung C, Tsai Y, Chang Y, Wu C, et al. Effects of Acute Exercise on Resting EEG in Children with
Attention-Deficit/Hyperactivity Disorder. Child Psychiatry Hum Dev 2018 Dec;49(6):993-1002. [doi:
10.1007/s10578-018-0813-9] [Medline: 29872997]
25. Hermens DF, Williams LM, Clarke S, Kohn M, Cooper N, Gordon E. Responses to methylphenidate in adolescent AD/HD:
evidence from concurrently recorded autonomic (EDA) and central (EEG and ERP) measures. Int J Psychophysiol 2005
Oct;58(1):21-33. [doi: 10.1016/j.ijpsycho.2005.03.006] [Medline: 15936104]

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 14


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

26. Kamida A, Shimabayashi K, Oguri M, Takamori T, Ueda N, Koyanagi Y, et al. EEG power spectrum analysis in children
with ADHD. Yonago Acta Med 2016 Jul;59(2):169-173 [FREE Full text] [Medline: 27493489]
27. Markovska-Simoska S, Pop-Jordanova N. Quantitative EEG in children and adults with attention deficit hyperactivity
disorder: comparison of absolute and relative power spectra and theta/beta ratio. Clin EEG Neurosci 2017 Jan;48(1):20-32.
[doi: 10.1177/1550059416643824] [Medline: 27170672]
28. Clarke AR, Barry RJ, Johnstone SJ, McCarthy R, Selikowitz M. EEG development in attention deficit hyperactivity disorder:
from child to adult. Clin Neurophysiol 2019 Aug;130(8):1256-1262. [doi: 10.1016/j.clinph.2019.05.001] [Medline: 31163371]
29. de Pasquale F, Della Penna S, Snyder AZ, Lewis C, Mantini D, Marzetti L, et al. Temporal dynamics of spontaneous MEG
activity in brain networks. Proc Natl Acad Sci U S A 2010 Mar 30;107(13):6040-6045 [FREE Full text] [doi:
10.1073/pnas.0913863107] [Medline: 20304792]
30. Brookes MJ, Woolrich M, Luckhoo H, Price D, Hale JR, Stephenson MC, et al. Investigating the electrophysiological basis
of resting state networks using magnetoencephalography. Proc Natl Acad Sci U S A 2011 Oct 04;108(40):16783-16788
[FREE Full text] [doi: 10.1073/pnas.1112685108] [Medline: 21930901]
31. Tang W, Liu H, Douw L, Kramer MA, Eden UT, Hämäläinen MS, et al. Dynamic connectivity modulates local activity in
the core regions of the default-mode network. Proc Natl Acad Sci USA 2017 Aug 21;114(36):9713-9718 [FREE Full text]
[doi: 10.1073/pnas.1702027114]
32. Hacker CD, Snyder AZ, Pahwa M, Corbetta M, Leuthardt EC. Frequency-specific electrophysiologic correlates of resting
state fMRI networks. Neuroimage 2017 Apr 01;149:446-457 [FREE Full text] [doi: 10.1016/j.neuroimage.2017.01.054]
[Medline: 28159686]
33. Samogin J, Liu Q, Marino M, Wenderoth N, Mantini D. Shared and connection-specific intrinsic interactions in the default
mode network. Neuroimage 2019 Oct 15;200:474-481 [FREE Full text] [doi: 10.1016/j.neuroimage.2019.07.007] [Medline:
31280013]
34. Greicius MD, Krasnow B, Reiss AL, Menon V. Functional connectivity in the resting brain: a network analysis of the
default mode hypothesis. Proc Natl Acad Sci U S A 2003 Jan 07;100(1):253-258 [FREE Full text] [doi:
10.1073/pnas.0135058100] [Medline: 12506194]
35. Greicius MD, Supekar K, Menon V, Dougherty RF. Resting-state functional connectivity reflects structural connectivity
in the default mode network. Cereb Cortex 2009 Jan;19(1):72-78 [FREE Full text] [doi: 10.1093/cercor/bhn059] [Medline:
18403396]
36. Catalá-López F, Hutton B, Núñez-Beltrán A, Page MJ, Ridao M, Macías Saint-Gerons D, et al. The pharmacological and
non-pharmacological treatment of attention deficit hyperactivity disorder in children and adolescents: a systematic review
with network meta-analyses of randomised trials. PLoS One 2017;12(7):e0180355 [FREE Full text] [doi:
10.1371/journal.pone.0180355] [Medline: 28700715]
37. Cortese S, Ferrin M, Brandeis D, Buitelaar J, Daley D, Dittmann RW, European ADHD Guidelines Group (EAGG).
Cognitive training for attention-deficit/hyperactivity disorder: meta-analysis of clinical and neuropsychological outcomes
from randomized controlled trials. J Am Acad Child Adolesc Psychiatry 2015 Mar;54(3):164-174 [FREE Full text] [doi:
10.1016/j.jaac.2014.12.010] [Medline: 25721181]
38. Tamm L, Epstein JN, Peugh JL, Nakonezny PA, Hughes CW. Preliminary data suggesting the efficacy of attention training
for school-aged children with ADHD. Dev Cogn Neurosci 2013 May;4:16-28 [FREE Full text] [doi:
10.1016/j.dcn.2012.11.004] [Medline: 23219490]
39. Tamm L, Nakonezny PA. Metacognitive executive function training for young children with ADHD: a proof-of-concept
study. Atten Defic Hyperact Disord 2015 Oct;7(3):183-190 [FREE Full text] [doi: 10.1007/s12402-014-0162-x] [Medline:
25559877]
40. Spencer-Smith M, Klingberg T. Benefits of a working memory training program for inattention in daily life: a systematic
review and meta-analysis. PLoS One 2016;11(11):e0167373 [FREE Full text] [doi: 10.1371/journal.pone.0167373] [Medline:
27875585]
41. Davis NO, Bower J, Kollins SH. Proof-of-concept study of an at-home, engaging, digital intervention for pediatric ADHD.
PLoS One 2018;13(1):e0189749 [FREE Full text] [doi: 10.1371/journal.pone.0189749] [Medline: 29324745]
42. Kollins SH, DeLoss DJ, Cañadas E, Lutz J, Findling RL, Keefe RSE, et al. A novel digital intervention for actively reducing
severity of paediatric ADHD (STARS-ADHD): a randomised controlled trial. The Lancet Digital Health 2020
Apr;2(4):e168-e178. [doi: 10.1016/s2589-7500(20)30017-0]
43. Kitzbichler MG, Henson RNA, Smith ML, Nathan PJ, Bullmore ET. Cognitive effort drives workspace configuration of
human brain functional networks. Journal of Neuroscience 2011 Jun 01;31(22):8259-8270 [FREE Full text] [doi:
10.1523/jneurosci.0440-11.2011]
44. Fornito A, Harrison BJ, Zalesky A, Simons JS. Competitive and cooperative dynamics of large-scale brain functional
networks supporting recollection. Proc Natl Acad Sci U S A 2012 Jul 31;109(31):12788-12793 [FREE Full text] [doi:
10.1073/pnas.1204185109] [Medline: 22807481]
45. Arbabshirani MR, Havlicek M, Kiehl KA, Pearlson GD, Calhoun VD. Functional network connectivity during rest and
task conditions: a comparative study. Hum Brain Mapp 2013 Dec;34(11):2959-2971 [FREE Full text] [doi:
10.1002/hbm.22118] [Medline: 22736522]

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 15


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

46. Finc K, Bonna K, Lewandowska M, Wolak T, Nikadon J, Dreszer J, et al. Transition of the functional brain network related
to increasing cognitive demands. Hum Brain Mapp 2017 Jul;38(7):3659-3674 [FREE Full text] [doi: 10.1002/hbm.23621]
[Medline: 28432773]
47. Custers EJFM, Durning SJ, ten Cate O. Principles and Practice of Case-based Clinical Reasoning Education. Berlin:
Springer; 2018.
48. Marling CR, Petot GJ, Sterling LS. Integrating Case-Based and Rule-Based Reasoning to Meet Multiple Design Constraints.
Computational Intell 1999 Aug;15(3):308-332 [FREE Full text] [doi: 10.1111/0824-7935.00095]
49. Marling C, Whitehouse P. Case-based reasoning in the care of alzheimer's disease patients. In: David W. Aha, Ian Watson,
eds.Case-based reasoning research and development. ICCBR 2001. lecture notes in computer science. Berlin, Heidelberg:
Springer; 2001.
50. Reddy M, Pesl P, Xenou M, Toumazou C, Johnston D, Georgiou P, et al. Clinical safety and feasibility of the advanced
bolus calculator for type 1 diabetes based on case-based reasoning: A 6-week nonrandomized single-arm pilot study.
Diabetes Technol Ther 2016 Aug;18(8):487-493. [doi: 10.1089/dia.2015.0413] [Medline: 27196358]
51. Harrison V, Proudfoot J, Wee PP, Parker G, Pavlovic DH, Manicavasagar V. Mobile mental health: review of the emerging
field and proof of concept study. J Ment Health 2011 Dec;20(6):509-524. [doi: 10.3109/09638237.2011.608746] [Medline:
21988230]
52. Kendig CE. What is proof of concept research and how does it generate epistemic and ethical categories for future scientific
practice? Sci Eng Ethics 2016 Jun;22(3):735-753. [doi: 10.1007/s11948-015-9654-0] [Medline: 26009258]
53. Finc K, Bonna K, He X, Lydon-Staley DM, Kühn S, Duch W, et al. Author Correction: Dynamic reconfiguration of
functional brain networks during working memory training. Nat Commun 2020 Jul 30;11(1):3891 [FREE Full text] [doi:
10.1038/s41467-020-17747-8] [Medline: 32732869]
54. Mishra J, Anguera JA, Gazzaley A. Video games for neuro-cognitive optimization. Neuron 2016 Apr 20;90(2):214-218
[FREE Full text] [doi: 10.1016/j.neuron.2016.04.010] [Medline: 27100194]
55. Conners K. Conners Continuous Performance Test (Conners CPT 3) & Conners Continuous Auditory Test of Attention
(Conners CATA): Technical Manual. Toronto, Canada: Multi-Health Systems; 2014.
56. Losier BJ, McGrath PJ, Klein RM. Error patterns on the continuous performance test in non-medicated and medicated
samples of children with and without ADHD: a meta-analytic review. J Child Psychol Psychiatry 1996 Dec;37(8):971-987.
[doi: 10.1111/j.1469-7610.1996.tb01494.x] [Medline: 9119944]
57. Schulz KF, Altman DG, Moher D, CONSORT Group. CONSORT 2010 Statement: updated guidelines for reporting parallel
group randomised trials. BMC Med 2010 Mar 24;8:18 [FREE Full text] [doi: 10.1186/1741-7015-8-18] [Medline: 20334633]
58. Boutis K, Willan A. Intention-to-treat and per-protocol analysis. Canadian Medical Association Journal 2011 Apr
04;183(6):696 [FREE Full text] [doi: 10.1503/cmaj.111-2034]
59. Ferguson CJ. Is Psychological Research Really as Good as Medical Research? Effect Size Comparisons between Psychology
and Medicine. Review of General Psychology 2009 Jun 01;13(2):130-136. [doi: 10.1037/a0015103]
60. Button KS, Ioannidis JPA, Mokrysz C, Nosek BA, Flint J, Robinson ESJ, et al. Power failure: why small sample size
undermines the reliability of neuroscience. Nat Rev Neurosci 2013 May;14(5):365-376. [doi: 10.1038/nrn3475] [Medline:
23571845]
61. Zalesky A, Fornito A, Bullmore ET. Network-based statistic: identifying differences in brain networks. Neuroimage 2010
Dec;53(4):1197-1207 [FREE Full text] [doi: 10.1016/j.neuroimage.2010.06.041] [Medline: 20600983]
62. Fornito A, Zalesky A, Bullmore E. Fundamentals of Brain Network Analysis. Amsterdam: Elsevier; 2016.
63. Twisk J. Applied Longitudinal Data Analysis for Epidemiology. New York: Cambridge University Press; 2013.
64. Castellanos NP, Paúl N, Ordóñez VE, Demuynck O, Bajo R, Campo P, et al. Reorganization of functional connectivity as
a correlate of cognitive recovery in acquired brain injury. Brain 2010 Aug;133(Pt 8):2365-2381. [doi: 10.1093/brain/awq174]
[Medline: 20826433]
65. Looi CY, Duta M, Brem A, Huber S, Nuerk H, Cohen Kadosh R. Combining brain stimulation and video game to promote
long-term transfer of learning and cognitive enhancement. Sci Rep 2016 Mar 23;6:22003 [FREE Full text] [doi:
10.1038/srep22003] [Medline: 26902664]
66. Nyquist JB, Lappin JS, Zhang R, Tadin D. Perceptual training yields rapid improvements in visually impaired youth. Sci
Rep 2016 Nov 30;6:37431 [FREE Full text] [doi: 10.1038/srep37431] [Medline: 27901026]
67. Soveri A, Antfolk J, Karlsson L, Salo B, Laine M. Working memory training revisited: A multi-level meta-analysis of
n-back training studies. Psychon Bull Rev 2017 Aug;24(4):1077-1096. [doi: 10.3758/s13423-016-1217-0] [Medline:
28116702]
68. Wang L, Hsiao C, Huang Y, Chiang Y, Ree S, Chen Y, et al. Association of salivary dehydroepiandrosterone levels and
symptoms in patients with attention deficit hyperactivity disorder during six months of treatment with methylphenidate.
Psychoneuroendocrinology 2011 Oct;36(8):1209-1216. [doi: 10.1016/j.psyneuen.2011.02.014] [Medline: 21411231]
69. Lawlor-Savage L, Goghari VM. Dual N-back working memory training in healthy adults: a randomized comparison to
processing speed training. PLoS One 2016;11(4):e0151817 [FREE Full text] [doi: 10.1371/journal.pone.0151817] [Medline:
27043141]

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 16


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

70. Melby-Lervåg M, Redick TS, Hulme C. Working memory training does not improve performance on measures of intelligence
or other measures of "Far transfer":evidence from a meta-analytic review. Perspect Psychol Sci 2016 Jul;11(4):512-534
[FREE Full text] [doi: 10.1177/1745691616635612] [Medline: 27474138]
71. Lenartowicz A, Delorme A, Walshaw PD, Cho AL, Bilder RM, McGough JJ, et al. Elctroencephalography correlates of
spatial working memory deficits in attention-deficit/hyperactivity disorder: vigilance, encoding, and maintenance. Journal
of Neuroscience 2014 Jan 22;34(4):1171-1182 [FREE Full text] [doi: 10.1523/jneurosci.1765-13.2014]
72. Deiber M, Hasler R, Colin J, Dayer A, Aubry J, Baggio S, et al. Linking alpha oscillations, attention and inhibitory control
in adult ADHD with EEG neurofeedback. Neuroimage Clin 2020;25:102145 [FREE Full text] [doi:
10.1016/j.nicl.2019.102145] [Medline: 31911342]
73. Nunez PL. Toward a quantitative description of large-scale neocortical dynamic function and EEG. Behav Brain Sci 2000
Jul;23(3):371-98; discussion 399. [doi: 10.1017/s0140525x00003253] [Medline: 11301576]
74. Takahashi S, Ohmiya M, Honda S, Ni K. The KCNH3 inhibitor ASP2905 shows potential in the treatment of attention
deficit/hyperactivity disorder. PLoS One 2018;13(11):e0207750 [FREE Full text] [doi: 10.1371/journal.pone.0207750]
[Medline: 30462746]
75. Tian L, Jiang T, Liang M, Zang Y, He Y, Sui M, et al. Enhanced resting-state brain activities in ADHD patients: a fMRI
study. Brain Dev 2008 May;30(5):342-348. [doi: 10.1016/j.braindev.2007.10.005] [Medline: 18060712]
76. Sokunbi MO, Fung W, Sawlani V, Choppin S, Linden DE, Thome J. Resting state fMRI entropy probes complexity of
brain activity in adults with ADHD. Psychiatry Res 2013 Dec 30;214(3):341-348. [doi: 10.1016/j.pscychresns.2013.10.001]
[Medline: 24183857]
77. Zhao Q, Li H, Yu X, Huang F, Wang Y, Liu L, et al. Abnormal resting-state functional connectivity of insular subregions
and disrupted correlation with working memory in adults with attention deficit/hyperactivity disorder. Front Psychiatry
2017 Oct 11;8:200 [FREE Full text] [doi: 10.3389/fpsyt.2017.00200] [Medline: 29075206]
78. Tomasi D, Volkow ND. Abnormal functional connectivity in children with attention-deficit/hyperactivity disorder. Biol
Psychiatry 2012 Mar 01;71(5):443-450 [FREE Full text] [doi: 10.1016/j.biopsych.2011.11.003] [Medline: 22153589]
79. Sonuga-Barke EJS, Castellanos FX. Spontaneous attentional fluctuations in impaired states and pathological conditions: a
neurobiological hypothesis. Neurosci Biobehav Rev 2007;31(7):977-986. [doi: 10.1016/j.neubiorev.2007.02.005] [Medline:
17445893]
80. Greenberg R, Daniels S, Flanders W, Eley J, Boring J. Interpretation of epidemiologic literature. In: Medical Epidemiology.
New York: McGraw-Hill; 2001:175-188.
81. Wolraich ML, Greenhill LL, Pelham W, Swanson J, Wilens T, Palumbo D, et al. Randomized, controlled trial of oros
methylphenidate once a day in children with attention-deficit/hyperactivity disorder. Pediatrics 2001 Oct;108(4):883-892.
[doi: 10.1542/peds.108.4.883] [Medline: 11581440]
82. Shuai L, Daley D, Wang Y, Zhang J, Kong Y, Tan X, et al. Executive function training for children with attention deficit
hyperactivity disorder. Chin Med J (Engl) 2017 Mar 05;130(5):549-558 [FREE Full text] [doi: 10.4103/0366-6999.200541]
[Medline: 28229986]

Abbreviations
ADHD: attention deficit hyperactivity disorder
ADHD-C: attention deficit hyperactivity disorder combined type
AI: artificial intelligence
BRIEF: Behavior Rating Inventory of Executive Function
CBR: case-based reasoning
CONSORT: Consolidated Standards of Reporting Trials
CPT: continuous performance test
CPT-III: Conner continuous performance test
DMN: default mode network
DSM-5: Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition
EDAH: Evaluación del Trastorno por Deficit de Atención e Hiperactividad
MEG: magnetoencephalography
PPV: predictive positive value

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 17


(page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH Medina et al

Edited by R Kukafka; submitted 03.11.20; peer-reviewed by A Lenartowicz, B Nievas Soriano; comments to author 18.11.20; revised
version received 11.12.20; accepted 13.05.21; published 26.11.21
Please cite as:
Medina R, Bouhaben J, de Ramón I, Cuesta P, Antón-Toro L, Pacios J, Quintero J, Ramos-Quiroga JA, Maestú F
Electrophysiological Brain Changes Associated With Cognitive Improvement in a Pediatric Attention Deficit Hyperactivity Disorder
Digital Artificial Intelligence-Driven Intervention: Randomized Controlled Trial
J Med Internet Res 2021;23(11):e25466
URL: https://www.jmir.org/2021/11/e25466
doi: 10.2196/25466
PMID:

©Rafael Medina, Jaime Bouhaben, Ignacio de Ramón, Pablo Cuesta, Luis Antón-Toro, Javier Pacios, Javier Quintero, Josep
Antoni Ramos-Quiroga, Fernando Maestú. Originally published in the Journal of Medical Internet Research (https://www.jmir.org),
26.11.2021. This is an open-access article distributed under the terms of the Creative Commons Attribution License
(https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic
information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must
be included.

https://www.jmir.org/2021/11/e25466 J Med Internet Res 2021 | vol. 23 | iss. 11 | e25466 | p. 18


(page number not for citation purposes)
XSL• FO
RenderX

You might also like