If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address for correspondence: Jared McGuirt, PhD, MPH, Department of Nutrition, University of North Carolina Greensboro, 319 College Ave, 318 Stone Bldg, Greensboro, NC 27412
Examine the acceptability of an evidence-based, contextually tailored, virtual avatar coaching approach for nutrition education among adult-child dyads with low income.
Design
Structured observations and semistructured interviews.
Setting
Low-income communities in 2 midsized urban areas.
Participants
Fifteen African American adult and child (aged 5–10 years old) dyads recruited from community agencies.
Phenomenon of Interest
Structured observation during the use of the virtual avatar coaching program and qualitative feedback on program experience after using the program.
Analysis
Transcripts were independently coded by 2 trained coders. Content analysis was conducted to determine trends and patterns in responses, and salient quotes were extracted from the manuscripts.
Results
The program generated dialogue between the children and adults around preferences and dietary habits. Adults and children described the program as being helpful, informative, motivational, and fun. Adults and children liked that the avatar acknowledged them at a personal level, provided accountability, and was tailored to them and their environment. Children wanted the ability to customize the avatar, more gamification, and avatar demonstration of healthy behaviors. Children felt the program would improve their health behaviors and were interested in using it again.
Conclusions and Implications
This study highlights information that can be used to improve avatar-based programs for nutrition education with children and their adult caregivers who have low income. Future work should assess the impact of user customization experience, program impact on dietary behaviors, and resources required to scale and maintain the program with community agencies.
The high prevalence of childhood obesity in the US is commonly linked to the overconsumption of energy-dense foods, a lack of participation in moderate to vigorous physical activities, and a lack of access to quality nutrition education.
Desired behavioral change will only occur at meaningful levels if common barriers to the accessibility and adoption of obesity prevention programming are addressed.
Previous research has found that for families with low income, particularly those living in rural areas, there may be several factors impacting access to adequate nutrition education, including limited program availability, short program duration (which may be an insufficient intervention dose), transportation challenges, and family responsibilities that prevent or limit attendance.
Barriers and facilitators to initial and continued attendance at community-based lifestyle programmes among families of overweight and obese children: a systematic review.
What works in practice: user and provider perspectives on the acceptability, affordability, implementation, and impact of a family-based intervention for child overweight and obesity delivered at scale.
One approach to address these barriers is virtual reality (VR), which is an interactive computer-generated experience that takes place within a simulated environment. Virtual reality technology can be accessed with a computer program (eg, with 3-dimensional models or avatars) or mobile phones, with or without a head-mounted display.
This immersive environment can be like the real world, or it can be fantastical, creating an experience that is not always possible in ordinary physical reality. Virtual reality can increase access to knowledge and experiences, provide the flexibility needed to address the wide-ranging needs of a childhood obesity prevention program, and offer a technological aspect of the programming that could be enticing to many children and families.
Avatar coaching, in which an icon or figure represents a person or character, could be a particularly engaging way to use VR to provide evidence-based nutrition coaching in a fun, social, and interactive way.
found that current evidence supports that avatars may have a positive impact on weight loss achievement and improved motivation; however, the review only reported 6 studies, all of which focused on adults. There is also a lack of virtual experiences that incorporate specific social and environmental contexts (eg, objective or subjective food environment). It has been established in the scientific literature and in theories and models of health behavior that the context of the environment is influential to health behaviors.
Ver Ploeg M, Breneman V, Farrigan T, et al. Access to affordable and nutritious food-measuring and understanding food deserts and their consequences: report to congress.http://www.ers.usda.gov/publications/pub-details/?pubid=42729. Accessed July 10, 2020.
On the basis of these gaps, the purpose of this mixed-methods study was to examine the usability and acceptability of a virtual avatar coaching program for adult-child dyads with low income. The program was designed to be evidence-based, community context-driven, and accessed via computer (requiring no head-mounted device in consideration of resource constraints). This innovation is based on the concepts of Social Cognitive Theory
which assert that learning and increases in self-efficacy best occur in a social context with a dynamic and reciprocal interaction of the person (in this case, an avatar), environment (in this case, environmental cues), and behavior (in this case, dietary intake).
In addition, intrinsic and extrinsic sources (like an avatar) can influence motivation and engagement for behavior change, including increasing autonomy, competence, and relatedness (development of personal relationships, in this case, with an avatar).
Adult-child dyads were recruited via flyers, targeted emails, and intercept recruitment at community agencies, organizations, and community centers in 2 urban areas in North Carolina. Eligible participants were: (1) qualified as low-income (federal benefit eligible); (2) parents or caregivers (herein referred to as an adult) and a child (aged between 5 and 10 years); and (3) English speaking. Investigators reviewed study information with both the child and adult participants, and both the child and adult verbally consented to participation. The research was approved by the University of North Carolina Greensboro Institutional Review Board (approval no. 19-0617).
In summer 2019, trained research staff conducted a mixed-methods study that included 2 components: (1) in-person structured observations of the child-adult dyads going through the VR program; and (2) interviews with the child-adult dyads about their experiences with the VR program and the potential impact of the program on their dietary behaviors. Data collection occurred at various community sites for participant convenience and in an isolated room to maintain privacy during data collection. Adults first completed a short demographic survey that included questions about the age and gender of the child participating, race/ethnicity, gender of the caregiver, relationship to the child, and household access to technology. Race or ethnicity was self-reported by the adult caregivers of the children from a list including non-Hispanic White, non-Hispanic Black, Hispanic, Asian or Pacific Islander, Native American (including Alaskan), or other (specify). The child-adult dyad then completed the VR program together on a single computer, with synchronous structured observations by the research team, followed by interviews with the research team.
There were 2 sections to the computer program. The first section was an age-appropriate child-focused section for child and avatar interaction, and the child and adult completed it together (with the child leading the interaction). This section was followed by an adult-focused section, not requiring child involvement, for more in-depth and higher-level adult and avatar interaction. This section asked adults about their child's reported health behaviors derived from the child-focused section and about general age-appropriate health behavior recommendations for their child. The program experience lasted a total of 15–20 minutes (8–10 minutes for the dyad and the separate adult section). Adults and children were not given an orientation to the program so that the researchers could examine undirected real-world usability.
The virtual coach programming was developed with custom scripts in the Amazon Sumerian platform (Amazon Web Services, Inc, 2019) (see Figure). The avatar presented evidence-based nutrition education guidance on the basis of the Health Resources and Services Administration sponsored Bright Futures in Practice: Nutrition approach to providing health supervision guidelines for children of all ages.
The program incorporated age-specific information on growth, physical, social and emotional development, healthy lifestyles, common nutrition concerns, and interview questions around eating behaviors, food resources, and counseling.
FigureSelected scenes of the virtual reality avatar experience seen by participants.
During the dyad section, children were prompted by the avatar to respond to questions and topics through typing or clicking response options. The adult was encouraged to participate along with the child. Children were asked about typical snacking behaviors, including fruits, vegetables, sugar-sweetened beverages, and salty snack consumption. Using their reported behaviors, the child was given advice or encouragement by the avatar, on the basis of recommended dietary guidelines.
The children were prompted to create a virtual healthy snack, then provided with suggestions for how to make snacks healthy, delicious, and fun. The program also included community context-specific guidance based on the user's physical residential address (entered into the system at the beginning of the program) and user identified locations of typical snacking behaviors (identified during the conversation with the avatar). For instance, if the child lived in a US Department of Agriculture designated food desert, their environmental context was acknowledged, and they were given tips and suggestions on how to identify and select the limited healthier items that may exist in that environment.
In the adult-focused section, adults (with child participation optional) were prompted by the avatar to respond to questions about the same topics discussed with children. Adults responded by typing or clicking response options. After the selection of an option, specific guidance was provided. In addition, guidance was tailored on the basis of the child's body mass index (BMI) derived from height and weight data reported by the adult. If the child's BMI was considered out of the healthy range (eg, overweight [≥85th percentile]) according to Centers for Disease Control and Prevention age and sex-specific growth charts), adults received additional information on the child's diet and physical activity behaviors.
If the BMI was considered within the healthy range, adults were told that their child was normal weight and provided with ways to maintain healthy behaviors.
Structured Observations
For the structured observations, a laptop with the avatar program was placed in front of the child and adult, and they were instructed to go through the program together. Research staff observed both the child and adult going through the program using an observation form developed for the study to track and note the child and adult's experience (Supplementary Data). Researchers were trained to use the observation form using an established protocol and practiced using the form during trial observations. The data collection form had sections for the observer to take notes about the following events: (1) initial child and adult reaction to the program, (2) child and adult usability, (3) points of difficulty or confusion while using the program, (4) points of praise from the child and adult about the program, and (5) discussion during program experience and ending commentary. Participants were not instructed to speak aloud about usability or confusion to facilitate natural behavior and immersion within the program.
In-Depth Interviews
Following the structured observation, the child-adult dyads completed a joint, semistructured in-depth interview that enabled researchers to understand their experiences and gather feedback about the program. The interview was performed by members of the research team trained in qualitative methods and who had practiced conducting the interviews during trial runs. The children and adults were individually asked about their initial thoughts about the program, what they learned from the program (to determine the most salient information), what was missing from the program, what they liked most and liked least about the program, what they would do if they designed the program themselves, whether they would use the program again, if they thought the program would change their behaviors, thoughts on the avatar, and thoughts on the context-based guidance for their food and physical activity environment. The interviews lasted approximately 30 minutes, with each dyad receiving $50 for their participation.
Data Analysis
Both the structured observations and in-depth interviews were audio-recorded and transcribed verbatim by the research team (B.E. and L.M.). A detailed consensus codebook was developed by the research team after reviewing an initial set of 3 transcripts, with the generation of both inductive and deductive codes. The transcripts were then independently coded by 2 trained coders using Atlas.ti (version 8, Atlas.ti Scientific Software Development GmbH, 2020). Coders met to discuss coding discrepancies and came to a consensus on how to code segments of text. A content analysis was conducted to determine trends and patterns in responses, and salient quotes were extracted from the manuscripts.
RESULTS
Participant Characteristics
A summary of the characteristics of adult-child dyads can be found in Table 1. A total of 15 adult-child dyads (n = 30 total participants) were recruited over 2 months. All adults reported African American race, the average age of the child was 9 years (range, 5–10 years), and there was an almost even distribution by child gender. Less than half of adults reported that the household had access to a computer at home, but most reported access to a computer away from home or a smartphone.
Table 1Characteristics of Adult-Child Dyads (n = 15)
Children and adults were typically excited to talk with the avatar, were engaged with the program, and were attentive when first introduced to the avatar program. Children made statements such as “How does she know me?” and “Cool, she got my name right.”
Most children and adults were able to navigate the program successfully, though some of the younger children had challenges, including typing and using the scroll menu to select snack choices. Some adults had issues navigating between outside internet sites and getting back to the program.
The most common point of difficulty was the adult estimating child height and weight for BMI calculations (children did not see this portion of the program). In addition, a few adults had difficulty typing in their residential address so that the program could accurately determine the food environment context for the contextually tailored guidance from the avatar.
Children reported enjoying the program, with 1 child saying, “I want to do it again. That was fun!” They liked the interactive nature of the program. Children responded positively to the tailored communication from the avatar, especially when the avatar responded positively to the healthy snack the child had created. Adults often made comments like “That is true” or confirmed the communication from the avatar about healthy habits and their food environment context.
Adults and children had the most discussion during the section on the child's favorite snacks, followed by servings of fruits and vegetables (FV) consumed yesterday, total snacks consumed yesterday, and making a smoothie. Most commonly, adults and children were clarifying behaviors (eg, what the child had recently consumed or typical dietary habits), followed by discussing directions (eg, how to properly respond to the avatar's questions, encouraging the child to pay attention), dietary habits (eg, typical eating behaviors, new eating preferences), and accountability (eg, adult holding child accountable for dietary behaviors).
After the program, there was often discussion and confirmation of the information learned between the adult and child. An example statement from an adult to a child was, “Did you hear what she said about fruits and vegetables?” Others gave immediate feedback about the program, like “Great tool. Will there be a phone app?”
Interview Section
A summary of findings by topic area with illustrative quotes can be found in Table 2.
Table 2Summary of Findings of by Topic Area With Illustrative Quotes
Feedback Target
Illustrative Quotes
Initial feelings
Adult participant 53: “I like that it integrated all the good foods you are supposed to eat in a day and the amount of food and healthy snacks in the bad snacks you're supposed to have per week. Which we both don't do.” Child participant 34: “I liked the talking and stuff. She was talking about like junk food is not good for you...Yea {liked that it felt like they were talking to a person}, but she is not real right? Is she?... It was like, you type in something and she understands you.”
Learned from the program
Adult participant 25: “I told you it's a good program. Seriously...Yea, it teaches you a lot... ‘cause it's a lot of stuff I did not know, but I know now.” Adult participant 9: “I learned that he {their child} likes bananas and strawberries, but I didn't know that... now I know certain stuff he likes.” Child participant 31: “That eating junk food is not good for you.”
Liked most about the program
Adult participant 34: “I thought that was really neat {information tailored to them and their environment}. That stood out to me. We do need that. I would recommend that. It makes it a little bit easier too {to make better decisions}. Yea, on what to eat to stay healthy and fit.” Child participant 34: “Her hair. I liked the talking about how junk food is not good for you. It was like, you type in something and she understands you.”
Liked least about the program
Adult participant 25: “The music was irritating, ‘cause it's kind of throwing me off from trying to listen to her.” Child participant 34: “Yea, like short, shorten the talk. Don't keep talking to me before I write.”
Participant design suggestions
Adult participant 9: “If you pick a good smoothie and then it'll {avatar} tell you how to make it. Yeah, that would be really cool.” Child participant 9: “I would, I would customize the character...Yeah, get to customize the background and stuff…I want to do a little mini-game, show you how to make this, make fruits and stuff.”
Potential future use
Adult participant 53: “Yea I would definitely use it again, just because it not only just help kids, it just lets adults know that we are the leaders and the kids are going to follow after what we eat.” Child participant 34: “It's fun… I never like did anything like where you just type in, and she just talks.”
Avatar program influence on behaviors
Adult participant 5: “Yea. I would eat less snacks, like chips and stuff and soda.” Child participant 57: “Not to drink a lot of soda.”
Thoughts on the avatar
Adult participant 56: “I agree. I liked the interaction. The response was quick. Like there was not a glitch or freeze. It was there. Once you type it in like it's a {snaps fingers} a quick response.” Child participant 56: “I like how she was virtual and how she was the one asking questions and telling me about it.”
Thoughts on food and physical activity context guidance
Adult participant 23: “Well I know what I liked. It told me that there weren't a lot of healthy options in the area that I'm in. And that is true because it's more convenience stores and Walmart. And it's a lot easier to get to the convenience stores.” Child participant 53: “Yea. That's pretty cool.”
Cultural relevance
Adult participant 53: “That was my initial favorite part, seeing her, that she was Black. I thought she was relatable.” Child participant 25: “{Want more} ‘hood music.”
All children and adults reported liking the program. Children and adults stated that the program was helpful, informative, motivational, and fun. To describe the program, children and adults used the terms “cool” and “fun” but also noted “too much talking” and “slow.”
Children and adults reported that they liked the program because of the interaction, educational content, child focus, pictures/graphics, tailored advice, and relatability of the avatar. Both children and adults reported enjoyment interacting with the avatar, including the ability to dialogue back and forth and discuss dietary habits. Several adults mentioned that the avatar brought accountability by helping them acknowledge unhealthy behaviors for themselves or their children. Children liked that the avatar acknowledged them at a personal level, and adults were excited about the educational content and that the program held their child's attention.
When asked what they learned from the program, children and adults most often stated that they learned about healthy dietary/snacking recommendations, distinguishing between healthier vs less healthy foods, eating more fruit (children only), child's food preferences (adults only), new foods and new food preparation methods. The program also provided an opportunity for adults to learn more about their child's food preferences and for children and adults to discuss plans on the basis of this discovery, such as buying new types of FVs or making smoothies together.
Children and adults reported a few things that they did not like including the background music (which distracted from the avatar interaction), being asked their residential address (though they felt more comfortable after explanation of the ability to provide tailored responses), how the avatar pronounced their name, the amount of time the avatar spent talking, seeing pictures of food (making them hungry), and that a new tab would open with additional information (this made navigation back to the program confusing). One adult suggested including personalized suggestions for local community food resources, and another suggested a stronger emphasis on children listening to their parents. Overall, however, most children and adults felt that the program was not missing any important components.
Children and adults mentioned a desire for customization of scenes and avatars, gamifying/playing a mini-game, including more pictures, avatar role modeling (including shopping at a grocery store or making a smoothie), and improved music.
Almost all children and adults thought the program had the potential to change their health-related behaviors. Adults mentioned changing purchasing habits on the basis of increased knowledge of their child's food preferences, and both children and adults noted the program would help them to eat more FV, fewer unhealthy snack foods, and drink less soda. Participants who did not see the potential for using the program to improve health behaviors reported already having healthy eating behaviors or noted financial constraints impacting the ability to purchase fresh FVs.
The overwhelming majority of children and adults said they would use the program again because it was educational, fun, and increased accountability.
DISCUSSION
The results of this study indicate that a VR avatar-based coaching program shows promise to increase access to and extend the reach of nutrition education programs to children from low income backgrounds who are at risk for obesity. All children and adults reported liking the program and planned to use it in the future, as they found it fun, informational, and motivating. The personalized social aspect of the avatar experience was also appealing, as participants thought the avatar would reinforce guidance and provide support while acting as a cue to change health behaviors. These findings align with the theoretical constructs of Social Cognitive Theory and Self-Determination Theory used to inform the design of the program and indicate that the combined social dynamic of the avatar and environmental cues can potentially influence motivation and behavior.
The results of this study are aligned with similar research that demonstrated the potential of using avatar coaches with adolescents in clinical weight management settings. In focus groups with teenagers, LeRouge et al
found that teenagers thought that an avatar program would: (1) reinforce guidance and support, (2) fit within their lifestyle, and (3) help set future goals. The teenagers in that study also expressed that they would be more willing to accept personalized advice from a virtual agent, such as an avatar, especially when the avatar was viewed as an extension of themselves.
The sample from this study also expressed the desire to increase personalization by customizing the avatar's clothing, hair, and physical features, and background scenery, which aligns with findings from previous research on games for health and children.
When games are designed around identity, the players have an opportunity to become a game character (via an avatar and/or to form relationships and linkages with game characters), thus providing them with a sense of agency.
This game design feature is particularly appealing to children because it enables players to manage aspects of gameplay, such as the use of control mechanisms and/or influencing storylines. The child participants in this study also expressed a strong desire to make the experience more game-like and interactive, which aligns with the previously found effective game features of interactivity, feedback, and immersion.
A key finding in this study was the avatar's ability to spark dialogue between the children and adults around dietary habits and behavior. Children revealed preferences for snacks, which prompted adults to make plans to change eating behavior on the basis of this new knowledge and increase purchases of FVs. These findings indicate that the avatar coaching program may help increase communication between adults and children around dietary intake that may not organically take place at home. The findings from this study align with existing literature demonstrating improved outcomes in community-based nutrition education programs in which adults and children can learn together.
The effect of iCook 4-H, a childhood obesity prevention program, on blood pressure and quality of life in youth and adults: a randomized control trial.
also found that adult-child joint participation was a strong motivating factor of family enrollment and retention in community-based programs for overweight youth and their families.
The interviews confirmed findings from the structured observations and revealed important changes that could be made to the program. Children had some challenges with typing and scrolling, so programs geared at this age group might focus on point and click buttons. The ease of use will be important for the long-term use of the program by the children. In addition, adults were unable to estimate their child's height and weight accurately. This issue led to inaccurate BMI estimates and guidance from the avatar. Although avatars can serve as virtual educators and provide evidence-based guidance, weight-related guidance on the basis of BMI and other clinical indicators may be avoided unless reliable measures or objective assessments are available. For requested user input like addresses, more guidance would ensure the correct address is entered.
This study did have some limitations. The qualitative nature of this study and the use of a convenience sample limits transferability to other populations. Social desirability bias may have also influenced answers. The potential for coding bias also exists, although rigorous research methods were employed to minimize bias.
and included both male and female adults/caretakers. This study also used a structured and systematic qualitative methodological approach to identify common themes on the potential effectiveness of this approach to improving dietary decision making.
IMPLICATIONS FOR RESEARCH AND PRACTICE
This novel study provides context into the experiences and perceptions of a sample of minority children and their adult caregivers from low-income backgrounds regarding VR coaching approaches for nutrition education. The information in this manuscript provides critical feedback that can be used to improve avatar-based programs targeting dietary intake in children and could inform future VR-based weight management programs for youth at high-risk for obesity. Furthermore, the use of avatars in nutrition education programs geared toward children is a highly innovative and relatively new approach. Continued improvement in technology and programmable features will likely provide the opportunity to customize the features of the avatar in future iterations of the program, which may help increase reach and the motivation of minority, children from low-income environments at risk for obesity to participate.
Future research should examine the influence of increasing the opportunity to personalize the program through user customization of avatar characters and settings. Further exploration into the feasibility, barriers, and facilitators of its adoption within partner organizations is warranted. This program was developed to be a free resource for individuals participating in nutrition education programs and designed to be easily modified by someone without computer programming skills. However, further examination of the costs and resources for entities to provide access to, maintain, and update these types of programs could be completed. This VR technology could also be accessed using head-mounted devices that could increase how immersive the program is for the user (and therefore effectiveness
With the children and adults using the laptop together, there is a chance that the presence of the adults could have impacted this study's examination of children's interactions with the program, and the child's sense of immersion within the program, which could impact effectiveness.
Future research might consider examining the independent impact of the virtual avatar program on children without adults. Future work could focus on delivering these types of programs through mobile phone applications. Using a mobile smartphone approach may change the level of immersion, and the impact of this change could be examined. Given the overwhelmingly positive reaction and strong interest from child and adult participants in this study for the avatar coaching program, further research is warranted to assess and quantify the impact of this type of approach on short- and long-term dietary behaviors.
ACKNOWLEDGMENTS
Support for this article was provided by the Health Resources and Services Administration Maternal and Child Health Bureau's Grand Challenges for Preventing Childhood Obesity.
Barriers and facilitators to initial and continued attendance at community-based lifestyle programmes among families of overweight and obese children: a systematic review.
What works in practice: user and provider perspectives on the acceptability, affordability, implementation, and impact of a family-based intervention for child overweight and obesity delivered at scale.
Ver Ploeg M, Breneman V, Farrigan T, et al. Access to affordable and nutritious food-measuring and understanding food deserts and their consequences: report to congress.http://www.ers.usda.gov/publications/pub-details/?pubid=42729. Accessed July 10, 2020.
The effect of iCook 4-H, a childhood obesity prevention program, on blood pressure and quality of life in youth and adults: a randomized control trial.