Biofeedback-based Digital Games and Well-being in Childhood: A Systematic Review

Sample sizes of the 16 included studies ranged from 8 participants to 1,045 participants (M = 98.6; Md = 30.5). Participants ranged in age from 3 to 30 years old, with average sample ages ranging from 9 to 17 years old. In terms of gender, samples ranged from 13% to 100% female. Of note, only four studies included racial and/or ethnic identities of participants (Burkhart et al., 2018; Ducharme et al., 2021; Knox et al., 2011; Scholten et al., 2016).

The majority of studies that reported the geographic location of the sample were conducted in the U.S. (Burkhart et al., 2018; Ducharme et al., 2021; Fish, 2018; Kahn et al., 2013; Knox et al., 2011; Mannweiler et al., 2023; Rush et al., 2017; Wintner et al., 2022; Woodberry et al., 2021). Four studies were conducted in the Netherlands (Bossenbroek et al., 2020; Scholten et al., 2016; Schuurmans et al., 2015, 2018), and one study was conducted in each Australia (Amon & Campbell, 2008) and New Zealand (Thabrew et al., 2021). Lastly, five studies reported collecting their data within a hospital setting (Burkhart et al., 2018; Ducharme et al., 2021; Kahn et al., 2013; Thabrew et al., 2021; Woodberry et al., 2021). Three studies were conducted in school settings (Bossenbroek et al., 2020; Rush et al., 2017; Scholten et al., 2016), and an additional three studies were conducted in residential care settings (Jaramillo-Quintanar et al., 2020; Schuurmans et al., 2015, 2018). Two studies were conducted on university campuses (Fish, 2018; Knox et al., 2011). The final two studies were conducted in a community setting (Mannweiler et al., 2023) and online (Wintner et al., 2022). See Table 1.

Ten unique digital games were played across the sixteen studies. Four games were played via a computer, three games were played via a tablet or mobile device, two games were played via virtual reality (VR), and one game was played with a screen and handheld controller. Most studies incorporated HR or HRV biofeedback. Respiratory and SCL biofeedback were also used alone or in addition to HR or HRV biofeedback. See Table 2 for an overview of biofeedback mechanisms.

Three studies tasked participants with playing Wild Divine via a computer (Amon & Campbell, 2008; Burkhart et al., 2018; Knox et al., 2011). Depending on the version of the game, children either wore biofeedback sensors on their fingers (Amon & Campbell, 2008; Knox et al., 2011) or earlobes (Burkhart et al., 2018) to provide feedback via HR (Amon & Campbell, 2008), HRV (Burkhart et al., 2018; Knox et al., 2011), or SCL (Amon & Campbell, 2008; Knox et al., 2011). Wild Divine included games and activities in a fantasy land with an on-screen mentor that taught breathing techniques (Bell, 2003). For example, one activity allowed children to build a bridge in order to cross a path and enter the next activity if they reached a stable pattern of breathing; if the biofeedback equipment detected an increased or unstable HR, the player was hindered from moving onto the next activity (Bell, 2003).

An additional three studies asked participants to play Dojo, a videogame designed to reduce anxiety in adolescents (Scholten et al., 2016; Schuurmans et al., 2015, 2018). Dojo was played via a computer and included HRV biofeedback and emotion regulation training through gamified contexts (Schuurmans et al., 2015). After watching instructional videos on deep breathing techniques, progressive muscle relaxation, positive thinking, and guided imagery, players were tasked with immersive puzzles to practice these strategies, which increased in difficulty if the player’s HR increased (Schuurmans et al., 2015). For instance, the “Fear Dojo” taught deep breathing techniques before the player navigated a maze to evade an angry spirit, which was only possible if their HR remained low (Schuurmans et al., 2015). All three studies evaluating Dojo included a trained research assistant monitoring participants to answer any questions and maintain silence if played in a group setting (Scholten et al., 2016; Schuurmans et al., 2015, 2018). Two studies evaluated user experience and found high participant satisfaction with Dojo (Schuurmans et al., 2015, 2018).

Participants in one study utilized emWave, which was also played via a computer, while completing a curriculum designed to promote emotional awareness and regulation (Rush et al., 2017). Players wore a sensor on their earlobe or finger to monitor HR, HRV, and respiratory rate. First, players were taught slow and steady breathing by pacing their breathing with a bouncing ball while being provided with immediate feedback on their HR, HRV, and respiratory rate (Rush et al., 2017). Then, children played computer games and activities to practice these skills (Rush et al., 2017). For instance, they could play a racing game where the speed of the car increased if the player had a steady rhythmic breath and decreased if the player had a faster irregular breath or quicker HR (Rush et al., 2017).

Children in two studies played RAGE-Control via a computer, which used HR biofeedback and was designed as an add on to anger control therapy (Ducharme et al., 2021; Kahn et al., 2013). When playing RAGE-Control, the child controlled a spaceship that was being attacked by aliens or asteroids (Ducharme et al., 2021; Kahn et al., 2013). If the player’s HR was too high, their ship could only fire “blanks” at the aliens or asteroids (Ducharme et al., 2021; Kahn et al., 2013). Players were rewarded for maintaining a steady HR by having the ability to control their spaceship and protect it from attacks (Ducharme et al., 2021; Kahn et al., 2013). Kahn and colleagues (2013) found high treatment compliance and player satisfaction with RAGE-Control.

RAGE-Control was later developed into Mightier, which continued to integrate HR biofeedback but was played via a tablet. Mightier is commercially available and was used in an additional two studies (Mannweiler et al., 2023; Wintner et al., 2022). Players wore an armband that monitored their HR and communicated with the Mightier app via Bluetooth technology (Mannweiler et al., 2023; Wintner et al., 2022). Mightier games included a task and an inhibitor that was activated when the player’s HR reached 7 beats per minute above their unique rolling average threshold (Mannweiler et al., 2023; Wintner et al., 2022). For example, one game asked players to slice falling fruit and avoid slicing falling trash. If the inhibitor was activated, smoke rose from the bottom of the screen and obscured the falling objects. To remove the inhibitor, players could either independently regulate their HR or engage with a guided relaxation strategy available within the game (Wintner et al., 2022). Wintner and colleagues (2022) found that most children in their sample (68%) chose to engage with Mightier for 5 or more weeks.

One study tasked children with using The Pip, which is also commercially available on iOS and Android mobile devices. Players wore a sensor on their fingers, and The Pip integrated skin conductance feedback (Fish, 2018). Players could visualize their baseline stress levels as well as their stress levels while gaming (Fish, 2018). In this study, use of The Pip was combined with a stress tracker app to assess baseline stress levels, the Loom App to teach players how to lower stress via visual displays (e.g., the snowy winter graphics transitioned to a vibrant summer scene with lowered SCL), and the Relax and Race App in which players must remain calm for their dragon to fly faster and win the race (Fish, 2018). After the intervention, teachers of students who played The Pip reported that the sessions were helpful to and well-received by students (Fish, 2018).

Thabrew and colleagues (2021) co-designed and evaluated Starship Rescue, which was also played via a tablet. Starship Rescue was based upon a story of a space hospital caught in a vortex of anxiety (Thabrew et al., 2021). Players were provided psychoeducation on anxiety and taught somatic relaxation and cognitive strategies, which were practiced while wearing a HR monitor that provided feedback during relaxation exercises (Thabrew et al., 2021). During the study, parents were asked to help their child choose a reward for completing the intervention, asked to validate their child’s achievement, and emailed a summary of key learning points (Thabrew et al., 2021). Multiple participants reported that they enjoyed playing Starship Rescue during the study (Thabrew et al., 2021).

Two studies utilized digital games with VR components. Jaramillo-Quintanar and colleagues (2020) utlized i-Care in their study, which was designed to manage anxiety. During play, biosensors were worn on the finger to monitor oxygen saturation and HR as proxies for feelings of anxiety (Jaramillo-Quintanar et al., 2020). First, a baseline assessment was completed to assess the players unique physiological parameters (Jaramillo-Quintanar et al., 2020). Next, an instructor taught the player diaphramatic breathing and guided imagery using visual displays of the player’s real time oxygen saturation and HR (Jaramillo-Quintanar et al., 2020). Finally, the player chose a character and designed a virtual recreational space in which they could relax (Jaramillo-Quintanar et al., 2020). During play, physiological parameters were monitored but did not appear on the screen (Jaramillo-Quintanar et al., 2020). Secondly, Deep was also a VR biofeedback game designed to manage anxiety and was used in an additional study (Bossenbroek et al., 2020). Players wore a belt with a stretch sensor to monitor their breathing while playing, and their respiration rate controlled their movement through an underwater fantasy world (van, 2016). Slower and deeper breaths allowed for better movement abilities around the virtual world, and there were no in game goals to achieve (Bossenbroek et al., 2020).

Lastly, Woodbury and colleagues (2021) utilized CALMS, which was designed to facilitate stress management and interpersonal skills within families. During the intervention, two players used handheld controllers to move spaceships throughout a screen (Woodberry et al., 2021). The goal was to deflect asteroids, which was only possible when both players’ HRs remained below their unique threshold (Woodberry et al., 2021). Researchers intermixed verbal reviews of players’ emotion regulation skills, interpersonal dynamics, and play strategies (Woodberry et al., 2021). After the intervention, children and their parents reported satisfaction with CALMS (Woodberry et al., 2021).

Most studies evaluated internalizing symptoms or externalizing behaviors associated with mental health disorders (Amon & Campbell, 2008; Bossenbroek et al., 2020; Burkhart et al., 2018; Ducharme et al., 2021; Fish, 2018; Jaramillo-Quintanar et al., 2020; Kahn et al., 2013; Knox et al., 2011; Mannweiler et al., 2023; Rush et al., 2017; Scholten et al., 2016; Schuurmans et al., 2015, 2018; Thabrew et al., 2021; Woodberry et al., 2021). Two studies included a measure of specific skills or mechanisms of action that can contribute to the development of mental health disorders, such as emotion regulation (Mannweiler et al., 2023; Wintner et al., 2022), and four studies assessed psychopathology or functioning more generally (Amon & Campbell, 2008; Ducharme et al., 2021; Mannweiler et al., 2023; Woodberry et al., 2021. Of note, many studies included more than one outcome measure (N = 10; Amon & Campbell, 2008; Bossenbroek et al., 2020; Ducharme et al., 2021; Kahn et al., 2013; Knox et al., 2011; Mannweiler et al., 2023; Rush et al., 2017; Schuurmans et al., 2015, 2018; Woodberry et al., 2021). See Table 2.

The most common outcome measures evaluated internalizing symptoms. Seven studies evaluated 8 outcomes of anxiety or distress (Bossenbroek et al., 2020; Burkhart et al., 2018; Knox et al., 2011; Scholten et al., 2016; Schuurmans et al., 2015, 2018; Thabrew et al., 2021). Of these 8 anxiety-related outcomes, 7 detected statistically signficant decreases in anxiety or differences between groups, with effect sizes ranging from small to large (Bossenbroek et al., 2020; Burkhart et al., 2018; Knox et al., 2011; Scholten et al., 2016; Schuurmans et al., 2015, 2018; Thabrew et al., 2021). Four studies included an outcome related to stress or relaxation (Fish, 2018; Jaramillo-Quintanar et al., 2020; Mannweiler et al., 2023; Woodberry et al., 2021); two of which detected statistically significant improvement or differences between groups with small to large effect sizes (Mannweiler et al., 2023; Woodberry et al., 2021). Two studies included a measure of anger (Ducharme et al., 2021; Kahn et al., 2013). Kahn and colleagues (2013) detected statistically signifcant differences between groups but an effect size was not reported and not able to be determined. Lastly, one study assessed symptoms of depression and found statistically signficant differences between groups with a large effect size (Knox et al., 2011).

Six studies included a total of eight outcomes assessing externalizing behaviors (Amon & Campbell, 2008; Bossenbroek et al., 2020; Ducharme et al., 2021; Rush et al., 2017; Schuurmans et al., 2015, 2018). Four of these six studies detected statistically signficant decreases in behaviors or differences between groups (Ducharme et al., 2021; Rush et al., 2017; Schuurmans et al., 2015, 2018). Effect sizes ranged from negligible to large. Amon and Campbell (2008) measured symptoms of attention-deficit/hyperactivity disorder (ADHD; e.g., inatttention, hyperacticity, impulsivity) and found statistically signficant differences between groups with a large effect size. Last of all, Ducharme and colleagues (2021) assessed overt aggression and detected statistically signficant differences between groups with a large effect size.

Two studies included a measure of emotion regulation, which evaluated specific skills or mechanisms of action that prevent mental health disorders or signify mental health challenges (Mannweiler et al., 2023; Wintner et al., 2022). Mannweiler and colleagues (2023) found statistically significant differences in emotion regulation by treatment group with a large effect size. Wintner and colleagues (2022) also found signficant improvement in emotion regulation skills but with a small effect size.

Lastly, a quarter of the included studies assessed psychopathology or functioning more generally (Amon & Campbell, 2008; Ducharme et al., 2021; Mannweiler et al., 2023; Woodberry et al., 2021); three studies found statistcially signficant improvements or differences between groups (Amon & Campbell, 2008; Ducharme et al., 2021; Woodberry et al., 2021). Amon and Campbell (2008) found signficant differences in social, emotional, and behavioral functioning by group with a large effect size. Similarly, Ducharme and colleages (2021) identified signficant differences in psychopathology by group with a large effect size. Lastly, a feasibilty trial detected signficant improvements in social functioning with a small to medium effect size (Woodberry et al., 2021).

Sample sizes of the 7 studies with generalizable results ranged from 24 to 138 participants (M = 52; Md = 37), participant ages ranged from 7 to 17 years old, and gender ranged from 13% to 100% female. Studies were conducted in the U.S. or the Netherlands within hospital, school, university, residential care, and community settings. All 7 studies utilized a form of cardiac biofeedback; four studies utilized HR biofeedback (Ducharme et al., 2021; Kahn et al., 2013; Mannweiler et al., 2023; Rush et al., 2017), and four studies utilized HRV biofeedback (Knox et al., 2011; Rush et al., 2017; Scholten et al., 2016; Schuurmans et al., 2018). One study also used SCL biofeedback (Knox et al., 2011), and one study also used respiratory biofeedback (Rush et al., 2017). Participants played RAGE-Control (Ducharme et al., 2021; Kahn et al., 2013), Wild Divine (Knox et al., 2011), Mightier (Mannweiler et al., 2023), emWave (Rush et al., 2017), or Dojo (Scholten et al., 2016; Schuurmans et al., 2018).

Six of the studies with generalizable results included an outcome measure of internalizing symptoms. Three studies evaluated anxiety symptoms; two of which found significant decreases in anxiety symptoms with large effect sizes after playing digital games that incorporated HRV biofeedback (Wild Divine or Dojo) over 8 sessions (Knox et al., 2011; Schuurmans et al., 2018). Knox and colleagues (2011) evaluated the impact of Wild Divine in a sample of children with elevated anxiety. When controlling for pre-intervention scores, they found significantly lower levels of reported anxiety (and depression) in the group who played 8 sessions of Wild Divine than in the control group (Knox et al., 2011). Similarly, Schuurmans and colleagues (2018) evaluated the addition of Dojo to usual treatment at a residential care facility. After 8 30-minute sessions, both children and their mentors reported significant reductions in anxiety symptoms with a large effect size (Schuurmans et al., 2018). Alternatively, Scholten and colleagues (2016) evaluated Dojo with adolescents who reported elevated anxiety levels on a school-wide screener. After six 60-minute sessions, Dojo and the control game (Rayman 2: The Great Escape) both resulted in reductions of anxiety (Scholten et al., 2016), however the treatment group did have steeper decreases in personalized anxiety symptoms than the control group (Scholten et al., 2016). The differing results between these studies can possibly be explained by the differences in number of sessions or the composition of study participants. The two studies with large effect sizes recruited participants who reported elevated symptoms of anxiety (Knox et al., 2011; Schuurmans et al., 2018) or externalizing behaviors (Schuurmans et al., 2018) and included 8 treatment sessions. Alternatively, while Scholten and colleagues (2016) required participants to have elevated anxiety symptoms, they also excluded participants who were already receiving mental health care which may have resulted in a population with less severe psychopathology. In addition, participants only engaged in 6 sessions (Scholten et al., 2016).

An addditional two studies evaluated the ability of RAGE-Control to reduce anger via HR biofeedback (Ducharme et al., 2021; Kahn et al., 2013). Kahn and colleagues (2013) compared the impact of RAGE-Control to treatment as usual within an inpatient psychiatry unit and found significantly greater reductions in reported feelings of anger for children who played RAGE-Control for 5 30-minute sessions than children in the control group who completed treatment as usual (Kahn et al., 2013). Effect sizes were not reported and unable to be calculated. Similarly, Ducharme and colleagues (2021) compared anger control training with and without the addition of RAGE-Control. All children who received anger control training reported reduced experiences of angry feelings, so there were not signficant group differences (Ducharme at al. (2021). Lastly, Mannweiler and colleagues (2023) evaluated Mightier, an updated version of RAGE-Control, during summer programming within a community setting. Parents of children who played 6 30-minute sessions of Mightier, alongside biweekly social-emotional learning groups, reported significantly greater decreases in parenting-related stress after the intervention than parents of children in the control group (Mannweiler et al., 2023).

Three of the studies with generalizable results evaluated externalizing behaviors (Ducharme et al., 2021; Rush et al., 2017; Schuurmans et al., 2018). In addition to anger, Ducharme and colleagues (2021) examined overt aggression and oppositional behaviors before and after anger control training with and without RAGE-Control, which utilizes HR biofeedback (Ducharme at al. (2021). They found significantly greater pre-post decreases in parent-reported aggression and behaviors in the group that played RAGE-Control compared to the group that did not play RAGE-Control, with large and medium effect sizes, respectively (Ducharme et al., 2021). Rush et al. (2017) evaluated the impact of emWave, which utilizes HR, HRV, and respiratory biofeedback, in addition to a weekly mindfulness curriculum on behaviors within special education classrooms. They found significant decreases in off-task behaviors (e.g., turning around in one’s chair, speaking about unrelated topics, looking out the window) with a large effect size after approximately 12 10-minute sessions (Rush et al., 2017). Lastly, Schuurmans and colleagues (2018) examined externalizing behaviors, in addition to anxiety, at a residential care facility. After 8 30-minute sessions of Dojo in addition to usual treatment they identified a significant decrease in self-reported externalizing behaviors and a non-significant change in mentor-reported externalizing behaviors compared to children who did not play Dojo, with medium and small effect sizes, respectively (Schuurmans et al., 2018).

In addition to parent stress, Mannweiler and colleagues (2023) evaluated emotion regulation before and after playing 6 30-minute sessions of Mightier, which utilizes HR biofeedback, at a diverse communty summer camp. Parents of children who played Mightier in addition to biweekly social-emotional learning groups reported significantly greater increases in their children’s emotion regulation skills after the intervention, with a large effect size, compared to parents of children in the control group (Mannweiler et al., 2023).

Last of all, two of the studies with generalizable results explored psychopathology or functioning more generally (Ducharme et al., 2021; Mannweiler et al., 2023). In addition to anger, overt aggression, and oppositional behaviors, Ducharme and colleagues (2021) evaluated clinician-rated severity of psychopathology (ranging from ‘Normal, not ill’ to ‘Extremely ill’) before and after anger control training with and without RAGE-Control, which utilizes HR biofeedback. They found significantly greater pre-post decreases in psychopathology in the group that played RAGE-Control compared to the group that did not play RAGE-Control with a large effect size (Ducharme at al. (2021). Mannweiler and colleagues (2023) examined behaviors and feelings more generally, which included both internalizing symptoms and externalizing behaviors. They did not find signficant differences between the treatment and control groups, which may be due to the utilization of a more universal population (Mannweiler et al., 2023).

In general, biofeedback-based digital games offer promise as both a treatment and preventative intervention for a variety of youth mental health difficulties in both universal and clinical populations. Given their digital and gamified components, biofeedback-based digital games also have potential to serve as an engaging, accessible, and scalable intervention. The included studies support biofeedback-based games as a standalone treatment, as an add-on to existing mental health services, and their administration in a variety of settings. However, biofeedback-based games are a relatively new contribution to the field. To date, there are only 7 studies, with relatively small sample sizes, that offer generalizable results. Researchers have yet to extensively compare the relative efficacy of types of biofeedback or gaming platforms in this context. Also of note, there is a dearth of evidence exploring the impact of biofeedback-based games with racially and ethnically diverse populations.

Technology literacy, access to digital devices, and access to the internet are often cited concerns regarding accessibility. Recent evidence suggests that the presence of technology in the lives of American youth is all but ubiquitous (e.g., Brodsky et al., 2021; Marci, 2022). Even when examining subgroups based on race and ethnicity, parental education, and household income, no group reported less than 94% of households with technology and internet access (National Center for Educational Statistics, 2023). It is important to note that not all digital mental health interventions are created with user-centered design principles, built for diverse audiences, or well-supported by empirical evidence. Digital interventions have the potential to revolutionize mental health care access, however consideration must be taken when adopting and assessing potential interventions, especially for vulnerable populations.

Lastly, this review also contained strengths and limitations due to its design. The authors followed strict PRISMA guidelines and included a high level of double coding which enhanced the rigor of the review. In addition, studies were only included if they were published in peer-reviewed journals, which potentially increased the quality of findings but may have limited the inclusion of relevant unpublished or non-peer reviewed studies. Studies were also only included if they were available in English, and relevant studies not available in English may have been excluded due to this decision.