Neurofeedback training (NFT) is considered a primary or supplementary treatment for a number of disorders, including attention-deficit/hyperactivity disorder (ADHD) [1-5], anxiety [6-9], and depression [7,8,10]. The American Academy of Pediatrics [11] provided a “level 1 best support” rating of NFT as a safe and effective evidence-based therapy for childhood ADHD. Nonetheless, several significant barriers prevent patients from receiving quality neurofeedback therapeutics; for example, electroencephalography (EEG) systems are expensive, complex, and often only accessible at health care clinics. A recent pilot study [12] provided encouraging evidence for the efficacy of therapist-guided NFT, suitable for remote home-based use. Findings showed improved ADHD symptomatology in a small cohort of children after 9 weeks of NFT. The system was designed as an affordable convenient wireless alternative to clinic-based EEG. NFT users regulate neural activity through operant conditioning, which can lead to morphological changes in the brain [13,14] and calmer, more focused cognitive, affective, and physical functioning. Currently, little is known about the effectiveness of NFT systems in the field [15]; therefore, this retrospective open-label pilot study offers real-world data supporting the efficacy of remote NFT in improving brain health.

Unlike standard EEG systems, Myndlift is an easy-to-use tool for patients and clinicians (Figure 1). While wearing the validated EEG headband (Muse; InteraXon [16,17]) containing four dry recording electrodes (ie, anterior frontal [AF] 7, AF8, temporal pole [TP] 9, and TP10), one ground electrode, and one auxiliary wet electrode, the patient trains with an Android or iOS app linked to the headset by Bluetooth, which delivers visual and auditory feedback during YouTube videos or specialized games. When patients’ brain waves are in the desired range, positive feedback is delivered. A therapist can set or adjust the training protocol and monitor progress remotely via a cloud-based web service. The device incorporates an app-based assessment, lasting approximately 40 minutes, completed prior to NFT (ie, baseline) and periodically over the intervention period for longitudinal tracking of improvement.

Real-world studies provide external validity and accurately represent the heterogeneity of a patient population [18]. From the app, real-world data were collected from more than 500 participants on outcome measures, including pre- and postintervention assessments of validated symptom questionnaires, a cognitive test of attention and executive functioning (ie, continuous performance task [CPT]), and resting EEG markers. An efficacious system could serve as a reliable, cost-effective solution for users. In-clinic NFT costs approximately US $150 to $200 per session, with a minimum of 30 to 40 sessions typically recommended. In contrast to a cost-per-session model, remote NFT could offer monthly charges, ranging from US $200 to $500.

Given the success of neurofeedback for child ADHD, more adults with ADHD are turning to NFT for treatment. Currently, 6.76% of adults worldwide—translating to 366.3 million people—are affected [19]. ADHD is commonly recognized as a hypoaroused brain state [20]. In recent years, EEG measures have provided supporting evidence for popular theoretical models of hypoactivation [21] related to core symptoms of hyperactivity, inattention, and impulsivity [22]. The hypoarousal state is best localized to frontal and posterior regions [23] (ie, neuroanatomical structures subserving attentional networks [24,25]). EEG patterns of ADHD in children are characterized by elevated low-frequency power (ie, primarily theta) and reduced relative high-frequency power (ie, alpha and beta) [23,26-28], or an elevated ratio of the two (ie, low to high frequency). The theta/beta ratio (TBR) is the most common form of NFT in treating ADHD [29,30]; however, inconsistencies in the literature suggest that TBR [31,32] may not be reliable as a diagnostic measure [33]. This may reflect EEG heterogeneity across ADHD-diagnosed individuals (eg, Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition [DSM-5] subtypes; psychiatric comorbidities; age; and sex) [34,35]. For instance, although theta and beta power differences are evident in child ADHD [31,36], a recent review [37] suggested that theta and alpha frequencies may be more reliable markers for adults. Notably, most adult studies emphasize group differences in alpha power during eyes-closed conditions [38-43], while more recent work has identified elevated theta and delta power in adults with ADHD [44,45]. Given this evidence, this study investigates whether TBR versus the delta/alpha ratio (DAR) or the theta/alpha ratio (TAR) are biomarkers for adult ADHD. Overall, the study evaluates evidence for improvement in mental health via symptom questionnaires, a CPT, and hypothesized EEG markers. Findings have implications for the benefits of NFT and efficacy of a remote home-based system.

Participants, 13 years of age or older, signed up through their clinician or a clinician suggested by Myndlift and completed NFT at home or in clinic in a clinical care context. Informed consent was provided through the app, allowing participants’ anonymized data to be used for research. Data were included for analysis if baseline (ie, preintervention) assessment was conducted after 7 or fewer NFT sessions (ie, attributed to in-app NFT tutorial). For analyses of improvement, postintervention sessions occurred 30 to 180 days after baseline with 20 or more NFT sessions completed [46]. An average of 1 or more NFT sessions per week was required for inclusion, given that effective neurofeedback requires consistency [47,48], irrespective of the neurofeedback protocol used. Data were collected via the app.

Procedures were reviewed by an independent Institutional Review Board (IRB)—Pearl IRB—who permitted IRB exemption for analyses of data previously collected and deidentified, following the guidelines of the Declaration of Helsinki.

Participants performed neurofeedback protocols (Multimedia Appendix 1) that were customized by their clinicians and consistent with current literature [49].

Data from 560 participants met the criteria for inclusion in the analysis. Depending on clinical considerations determined by their therapist, subsets of participants completed each symptom questionnaire, CPT, resting EEG, or any combination of the three. Table 1 gives sample characteristics for each assessment component, including the NFT protocols completed by 50% or more of each sample population (Multimedia Appendix 1).

Results for participants who completed symptom questionnaires (n=301) were separated into groups with abnormal and healthy scores (Table 2). Most participants engaged in NFT protocols to reduce theta (227/301, 75.4%) and enhance high beta (248/301, 82.4%), while many who completed the PHQ-9 (76/134, 56.7%) and the ASRS (59/112, 52.7%) also performed enhanced alpha, whereas children who completed the ADHD-RS-IV also often included enhanced low beta (21/27, 78%) and enhanced sensorimotor rhythm (SMR; 16/27, 59%). In the groups with abnormal results, all questionnaires had large effect sizes (d=0.99 to 2.41), while the effect sizes for groups with healthy results were large only for child and adult ADHD questionnaires. Improvement in the groups with abnormal results was statistically significant for all questionnaires, with the majority (30/56, 54% to 7/7, 100%) of users demonstrating clinically significant change (ie, ≥20%) [65,66]. The most prominent improvement was observed in participants with abnormal baseline anxiety or child ADHD scores. Nevertheless, ADHD-RS-IV findings are considered preliminary given the small sample size. Most participants (30/56, 54% to 7/7, 100%) in the groups with abnormal results shifted their values to healthy ranges at postintervention. Improvement of healthy participants was statistically significant for all questionnaires, with the majority (30/66, 45% to 14/20, 70%) demonstrating clinically significant change.

Participants completing CPT and ADHD questionnaires performed primarily reduced theta (76/99, 77%) and enhanced high beta (81/99, 90%) protocols. Most adults also performed enhanced alpha (54/90, 60%), whereas most children also performed enhanced low beta (9/9, 100%) and enhanced SMR (7/9, 78%). Results (n=99) for average RT, SDRT, omission errors, and commission errors were divided by abnormal versus healthy scores for child and adult ADHD combined (Table 3). The greatest improvement observed, irrespective of group (ie, abnormal and healthy ADHD ranges), was in SDRT (d=1.02 and d=1.24, respectively), where nearly half of the participants (42/99, 43%) demonstrated clinically significant improvement, as indicated by the RCI. Although average RTs improved comparably (42/99, 43%), differences between pre- and postintervention were significant only for the healthy results group (d=0.56). At least one-third of users improved in their commission errors (35/99, 35%) and omission errors (45/99, 45%) from pre- to postintervention. Results from a group (n=104) with unknown ADHD assignment were comparable to those of groups with abnormal and healthy results (Multimedia Appendix 3).

This retrospective study offers initial evidence of therapist-guided remote neurofeedback as an effective tool for reducing subjective symptoms, improving objective cognitive performance, and adaptively modifying EEG markers. Improvements in attention were evident in children and adults with ADHD, as well as healthy participants. Findings suggest that the TBR is not a reliable marker for adult ADHD, instead demonstrating alternative elevated slow/fast power ratios [37]. Moreover, we provide preliminary evidence for improvement (ie, reduced DAR) in adults with ADHD. These findings offer a promising use for remote NFT as a low-cost alternative to clinic-based EEG.

Based on real-world data, significant improvement was reported across standardized questionnaires. The greatest improvement was observed in participants with abnormal anxiety scores, where most received reduced theta, enhanced high beta, and enhanced alpha protocols. As anticipated, greater effect sizes were observed for participants with scores in the abnormal versus healthy ranges. Interestingly, healthy participants and those with ADHD, both children and adults, demonstrated significant improvement with large effect sizes after completing primarily reduced theta and enhanced high beta protocols, as well as adults who completed enhanced alpha protocols or children who completed reduced low beta and reduced SMR protocols. Consistent with the literature [33-35], our findings suggest that children and adults may benefit from unique NFT protocols to improve ADHD symptoms, although a larger sample is required to confirm preliminary ADHD-RS-IV results.

Apart from the child ADHD assessment, questionnaire analyses included large total sample numbers (ie, 112 to 263 participants), and after an average of 53 NFT sessions, 57% to 78% of the participants demonstrated significant improvement, depending on the questionnaire. Results were particularly impressive compared to other in-app mental health therapeutics [73-76], such as mobile-enabled text psychotherapy [77] or app-based cognitive behavioral therapy [78]. The majority (61%) of participants scoring in the abnormal ranges moved to the healthy results group over an average of approximately 3 months, a time frame costing less than US $1500 with Myndlift versus US $6000 to $8000 for traditional neurofeedback.

NFT led to greater consistency in response times on a response inhibition task for subjects scoring in healthy or abnormal ADHD ranges, agreeing with similar reports of subjects with ADHD [79,80]. In addition, the RCI demonstrated that approximately 50% of healthy participants improved their average response time, while similarly, participants in abnormal ranges reduced omission errors. Importantly, CPT findings agree with improved ADHD questionnaire scores, suggesting that NFT provides objective evidence of improved executive function, the primary cognitive domain impacted by attentional difficulties.

Resting EEG findings demonstrated that elevated DAR and TAR were indicative of adult ADHD at baseline. This translated to significantly higher levels of delta and lower levels of alpha, as previously reported in adults with ADHD [39-42,81,82]. Notably, Liechti and colleagues [35] reported high theta to be less consistent in adults than in children, and that ADHD versus healthy control classification improved having exploratorily included delta waves in the discriminant analysis. Adults with ADHD may present slower theta waves—bordering fast delta waves—than children, although further analysis is required. Together, findings are consistent with the cortical hypoarousal theory, where low-power fast oscillations accompany reduced self-control and executive functioning [83], and high-power slow oscillations are reported with decreased subcortical motivational drive [84]. Preliminary evidence for reduced DAR in adult ADHD from pre- to postassessment may reflect the improved ADHD symptoms and CPT measures, particularly given the success of protocols inhibiting slow oscillations and enhancing fast oscillation [1], and the high percentage of ADHD participants performing reduced theta (ie, slow) and enhanced alpha (ie, fast) protocols.

In contrast to our work and that of others, several groups reported high alpha power at baseline during eyes-closed conditions in adult ADHD populations [85,86], or rather, no difference across ADHD participants and healthy controls [87,88]. Importantly, variability across the adult ADHD literature may, in part, be due to the heterogeneity of ADHD [34,35] and differences in study designs, sample sizes, analyses, and EEG technology [89]. For example, Loo and colleagues [38] demonstrated that adults with ADHD combined-type (ie, symptoms of inattention and hyperactivity or impulsivity) present reduced alpha power globally, compared to ADHD inattentive-type or non-ADHD controls.

Study results are encouraging, but conclusions should be tempered by limitations, including small subgroup sample sizes and lack of control groups. Moreover, subjects may have received alternative treatment in parallel (eg, medication) that could influence symptom improvement as well as alter neuromarkers. For example, two studies administering stimulants (ie, methylphenidate or dexamphetamine) to treat symptoms of ADHD in adults demonstrated altered delta [90,91] and theta waves [90] posttreatment. No changes in alpha or beta waves were reported. Given the evidence in this study for altered delta and alpha waves in adults with abnormal ADHD scores, we would hypothesize that the mechanism of action for stimulants versus NFT may differ, resulting in influence over varied frequency bands. Moreover, as this population reflects real-world use, the likelihood of these two forms of treatment to have commenced simultaneously, for treating symptoms of depression, anxiety, and ADHD, would arguably be low. Those seeking treatment with remote neurofeedback most often do so to avoid taking pharmaceuticals [92,93] or, rather, to supplement their current treatment, which alone may not be sufficiently effective [94]. Frank H Duffy [95], a Harvard professor and pediatric neurologist, suggests that “if any medication had demonstrated such a wide spectrum of efficacy it would be universally accepted and widely used.” Further, controlled research studies will be required to facilitate comparison of neurofeedback efficacy with other interventions. Notwithstanding these limitations, the findings are essential as they reflect real-world benefits of remote neurofeedback to actual patients. Follow-up analyses will compare benefits across NFT protocols and will further evaluate the impact on resting EEG outcomes.

Preliminary findings from this retrospective pilot study demonstrate efficacy of remote NFT in improving mental health, particularly for individuals with symptoms of ADHD and anxiety, mainly through reduced theta, enhanced high beta, and enhanced alpha NFT protocols. Moreover, adult ADHD was distinguished from healthy individuals by elevated frontal DARs, where ratios were significantly reduced following NFT. The effectiveness of the system in a real-world population via remote use positions it as an affordable and accessible alternative to clinic-based systems.