A Fully Automated Conversational Artificial Intelligence for Weight Loss: Longitudinal Observational Study Among...

Area of Science:

• Digital Health
• Artificial Intelligence in Healthcare
• Behavioral Science

Background:

• Type 2 diabetes is a leading costly chronic disease in the US, with a high prevalence of prediabetes and obesity.
• Intensive in-person behavioral counseling aids weight loss and health behavior change but faces limitations due to provider shortages and cost.
• There is a need for scalable, cost-effective interventions to manage chronic diseases and improve health outcomes.

Purpose of the Study:

• To evaluate the effectiveness of a conversational artificial intelligence (AI) health coach (HCAI) in promoting weight loss and improving meal quality.
• To assess user acceptability and engagement with the AI health coach as a mobile health intervention.
• To inform future clinical trials and AI design for chronic disease management and prevention.

Main Methods:

• A longitudinal observational study involving overweight and obese participants using the Lark Weight Loss Health Coach AI (HCAI).
• Data collection included self-reported weight, meal quality, physical activity, and sleep, with some automatic detection via mobile phone.
• User engagement was measured by app usage duration and session frequency; acceptability was assessed via a 4-question in-app trust survey.

Main Results:

• Participants achieving engagement standards (N=70) showed an average weight loss of 2.38% of baseline weight over 15 weeks.
• Users averaged 103 sessions, and predictors of weight loss included duration of AI use, counseling sessions, and meals logged.
• Healthy meal consumption increased by 31%, with high user satisfaction (87/100) and a net promoter score of 47.

Conclusions:

• AI health coaching is a viable alternative to in-person interventions, demonstrating comparable weight loss and significant behavior change.
• The AI health coach achieved high user acceptability, suggesting its potential to increase access to health care.
• Further research and clinical trials are warranted to explore AI's role in telemedicine for managing weight, health behaviors, and user engagement.