Project Bettie - Chatbots in Healthcare: Case Study in Diabetes Prevention

Project Bettie | Chatbots in Healthcare: Case Study in Diabetes Prevention
BFA Student: Jenny Williams
Faculty Advisor: Aynne Valencia and Haakon Faste

1 in 3 young adults (ages 18-39) are estimated to be pre-diabetic in the state of California, 70% of which will develop type II diabetes within their lifetime. In order to sidestep the stigma associated with diabetes and harness the popularity of Facebook Messenger, I created a Facebook chatbot, named Bettie, to help facilitate the behavior changes necessary to prevent young adults from developing type II diabetes.

According to the Center for Disease Control and Prevention, achieving a 5 to 7 % weight loss and logging 150 weekly minutes of moderate physical activity can reduce the onset of type 2 diabetes in pre-diabetics by nearly 60%. Additionally, a 2002 Kaiser Permanente study showed that people who keep a food diary double their weight loss efforts. In speaking with physicians who work with diabetic patients, I found that patients who have a social support structure are more likely to be succeed in making sustainable positive lifestyle changes. Although there are many health intervention tools on the market today, they tend to be inaccessible to many young adults and are not sustainable over time.

In order to sidestep the stigma associated with diabetes and harness the popularity of Facebook Messenger, I created a Facebook chatbot, named Bettie, to help facilitate the behavior changes necessary to prevent young adults from developing type II diabetes.

Drawing from my previous experience in the healthcare field, and using key insights from my research, I created Bettie Bot to help young adults track their food throughout the day though a conversation.  Bettie also provides people daily challenges to help encourage new habits and facilitate learning. Additionally, Bettie answers questions about nutrition as well as diabetes.

Living within Bettie Bot was an A/B test that sent participants to interact with two extremely different bot “personalities”: a formal reminding bot, and a friendly never-reminding bot. This allowed me to use Bettie as a research tool to test the key mechanics within conversational UI, specifically voice branding, triggers, and motivation in a case study focused on healthcare interventions.