The Impact of an Artificial Intelligence Chatbot on Brazilian Adolescents' Body Image

Background

High prevalence of body and eating concerns in Brazilian populations is well-documented, with risk observed across the lifespan. With a population of 211 million people, prevalence rates of body dissatisfaction range between 26.6 - 56% and 10.7- 36% for adolescent girls and boys, respectively (1, 2, 3). Equally, an emerging proportion of adolescents report engaging in disordered eating (i.e., binge eating [10.4%], dietary restriction [8.7%]) and weight control behaviours (i.e., vomit and laxative use [.3 - 7.4%]; 4,5). The prevalence of these disordered attitudes and behaviours are mirrored in older populations, and have shown to manifest within family units (6). A meta-analysis into eating disorder prevalence among Brazilian university students (Mage = 21.6 years) reported 14.9% of students met criteria for an eating disorder, with cases highest among nutrition students (7). Meanwhile, Brazilian parents' perceptions of their child's weight during early adolescence (i.e., 11 years), has shown to predict higher weight gain and engagement in weight control behaviours during late-adolescence (i.e., 18 years; 6). Brazilian consumer trends reflect these attitudes and behaviours, with it being the leading country in diet pill consumption and aesthetic surgeries, and second in the world for total aesthetic procedures (i.e., surgical and non-surgical), with 2.27 million procedures conducted annually (8). Despite the scope of body and eating concerns, few evidence-based interventions have been developed and tested for Brazilian populations and even fewer are accessible, scalable or cost-effective.

It is widely accepted among health professionals, policy makers and governments that the global burden of mental health will never be successfully met by extant health professionals and services alone. Further, current intervention modalities (e.g., face-to-face psychotherapy) may be unnecessary or unsuitable for milder concerns, which may be responsive to lower intensity interventions that demand fewer resources. Harnessing technology to deliver evidence-based care is a key focus for researchers. Micro-interventions (brief, low intensity, self-administered interventions), offer an alternative to traditional, intense interventions that may be unsuitable for milder concerns. Body image micro-interventions have proven effective at providing immediate and short-term improvements in body image among women (9). To date, body image micro-interventions have been developed and evaluated among high-income, White, English-speaking samples, with little research exploring how this intervention model may cater to other cultures and countries. Given the prevalence of body image concerns among Brazilian adolescents, coupled with health care disparities, developing evidence-based micro-interventions for scalability is an imperative step in mental health research for Brazil.

The aim of the present study is to evaluate a chatbot that targets risk and protective factors for body image among Brazilian adolescents. Using a two-armed randomised controlled trial design (i.e., intervention vs. assessment only), the study will assess the chatbot's efficacy at eliciting immediate and short-term improvements in adolescents' state- and trait-based body image, affect and body image self-efficacy. This is the first artificial intelligence chatbot to target body image among adolescents, particularly in Brazil; thus, a secondary aim is to assess user's intervention engagement and adherence.

A set of hypotheses were formulated for overall chatbot efficacy (H1-H2), moderation effects within sub-samples (H3) and intervention engagement and adherence (H4):

H1: The chatbot was designed to provide immediate benefits to users. Therefore, it is anticipated that adolescents will experience improvements in state-based body satisfaction and affect at the time of engagement with the chatbot.

H2: Adolescents randomised into the intervention condition will experience greater improvements in trait-based body esteem, affect and body image self-efficacy at post-intervention, one-week and one-month follow-up following a 72-hour trial period, relative to the assessment only control condition.

H3: Intervention effects will be moderated by gender, and baseline levels of body esteem, affect and body image self-efficacy. Specifically, intervention effects will be greatest among:

• Girls
• Girls and boys with lower levels of trait body esteem
• Girls and boys with higher levels of trait negative affect
• Girls and boys with higher levels of body image self-efficacy

H4: With regards to user engagement and adherence, given the novelty of this intervention, analyses will be of an exploratory nature. It is, however, anticipated that greater engagement and adherence with the chatbot intervention will lead to greater improvements in trait-based outcomes.

Design

This is a two-arm randomised controlled trial, comparing an online intervention to an assessment only control condition. Those in the intervention condition will interact with an artificial intelligence chatbot over a 72-hour period, while participants in the control condition will receive care as usual. Assessments of primary and secondary trait outcomes will be conducted at baseline (T1), post-intervention (T2), one-week (T3) and one-month follow-up (T4). Participants in the intervention condition will also be assessed on state body satisfaction and affect immediately before and after engaging with an intervention technique embedded into the chatbot.

Intervention

The body image chatbot, Topity, is designed to target sociocultural risk and protective factors for body image using eight, brief therapeutic techniques derived from several evidence-based theories, including media literacy, cognitive behaviour theory and positive body image. The intervention aims to engage users in techniques that teach users how to 1) Critically analyse and evaluate media content to reduce vulnerability to negative media influences (i.e., media literacy theory; 10); 2) Identify and challenge unhelpful thinking styles and behaviours that perpetuate body image distress (i.e., cognitive behaviour theory; 11); and 3) Appreciating the features, functions and health of the body, beyond it's appearance (i.e., positive body image theory; 12).

Participants

This universal intervention was designed to provide whole population benefits, irrespective of an individual's level of risk. Therefore, increased risk for body or eating concerns was not an inclusion criterion for this trial. Eligible participants are those aged 13-18 years old; Portuguese speaking; Brazilian resident; and, have access to Facebook Messenger.

Recruitment and data collection procedures

A representative sample of 2,800 adolescents aged 13-18 will be recruited via a Brazilian research agency, using the agency's recruitment panels (e.g., via email) and UNICEF's online communication platforms (e.g., U-Report). Recruitment materials will include a link to a screener to determine participant eligibility.

Based on the screening results, participants will receive an information pack to be read by themselves and their parent or guardian. Parents will be invited to provide consent by uploading a signed consent to an online, confidential portal. Following parent consent, participants will complete the online self-report assessment (baseline; T1), which will commence with adolescents providing informed assent, by clicking a checkbox. Those who do no assent will not proceed with the trial. At completion of the baseline survey, adolescents will be randomised into one of two conditions: intervention or assessment only. Those in the intervention condition will receive a link to the chatbot via email 24-hours after completing the baseline assessment. Those in the assessment only condition will be advised of their opportunity to engage with the chatbot at study completion.

Adolescents will be invited and encouraged to engage with the chatbot, as much as possible, over a 72-hour period. During this period, participants will have the opportunity to complete and re-engage with eight intervention techniques. Participants will be assessed on state body satisfaction and affect immediately after engaging with a technique. At the completion of the 72-hour period, both conditions will be assessed on primary and secondary trait outcomes (post-intervention; T2). Those in the intervention condition will be encouraged to continue using the chatbot. One-week (T3) and one-month (T4) following the intervention period, participants will be assessed on primary and secondary trait outcomes. At the completion of the one-month follow-up, all participants will receive a debrief form, outlining the study aims and objectives, and additional resources for body image and eating concerns. Those in the assessment only condition will be invited to engage with the chatbot; however, their engagement will not be monitored or assessed.