Skip to content

LiveBetterSA: Comparing AI-Enhanced Lifestyle Coaching Using a Generative Large Language Model vs. Traditional Telephone Coaching for Adults

LiveBetterSA: A Non-Inferiority Type 2 Hybrid Implementation-Effectiveness Randomised Controlled Trial Comparing AI-Enhanced Lifestyle Coaching Using a Generative Large Language Model vs. Traditional Telephone Coaching for Adults

Status
Not yet recruiting
Phases
Unknown
Study type
Interventional
Source
ANZCTR
Registry ID
ACTRN12625000412437
Acronym
LBSA
Enrollment
492
Registered
2025-05-06
Start date
2026-01-05
Completion date
2027-01-04
Last updated
2025-10-13

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

None listed

Brief summary

The LiveBetterSA trial has been designed to compare the effectiveness of artificial intelligence (AI) powered digital health coaching to traditional telephone-based health coaching in supporting healthy lifestyle changes. The study aims to evaluate whether the digital health coaching program is as effective as telephone-based health coaching on weight management at 6 months. Participants will be randomly assigned to receive either digital health coaching or telephone health coaching over six months. The study will also examine the effects of digital health coaching on secondary outcomes—including waist circumference, physical activity levels, dietary habits, sleep, and quality of life, compared to traditional telephone-based coaching. In addition, the study will assess key implementation outcomes including participant engagement and program acceptability. Finally, a health economic analysis will be conducted to assess whether the digital health coaching model offers a cost-effective alternative to traditional telephone coaching. This study represents the first large-scale evaluation of an AI-powered digital health coaching model for preventive health. If successful, it could offer a scalable, sustainable approach to improving population health in Australia.

Interventions

Participants in the digital health coaching group will engage with a digital health coach for six months via the LiveBetterSA responsive web application. The digital health coaching system, developed using OpenAI's large language model and customised by the University of South Australia and AndMine Pty Ltd, will provide personalised onboarding, guided goal setting, and feedback. The system will be accessible 24/7, allowing participants to interact with the digital health coach at their convenien

Participants in the digital health coaching group will engage with a digital health coach for six months via the LiveBetterSA responsive web application. The digital health coaching system, developed using OpenAI's large language model and customised by the University of South Australia and AndMine Pty Ltd, will provide personalised onboarding, guided goal setting, and feedback. The system will be accessible 24/7, allowing participants to interact with the digital health coach at their convenience. Optional Fitbit integration will enable automated step tracking. Participants will receive digital scales and a measuring tape for self-monitoring and outcome reporting. Participants are encouraged to engage with the platform daily, including asking for advice, responding to coaching prompts, reviewing goals, and receiving personalised feedback. The goal progress check-in schedule is customisable, selected by participants during onboarding, with options ranging from daily, every other day, twice weekly, weekly, or every two weeks. Personalisation occurs through participant-entered data collected during onboarding, including demographic information, anthropometric measurements, health history, physical activity patterns, dietary habits, eating behaviours, and personal motivations. This data informs tailored goal recommendations, feedback pathways, and support strategies, dynamically adjusted as participants engage with the platform over time. The guidance provided by the digital health coach is based on comprehensive training content sourced from evidence-based national and international guidelines, educational materials, and reputable health organisations. Key sources informing the AI's content include the Australian Dietary Guidelines (NHMRC) and Australian Physical Activity Guidelines (Department of Health and Aged Care), World Health Organization (WHO) guidelines, as well as educational resources from Diabetes Australia, the Heart Foundation, Cancer Council, Healthy Bones Australia, Exercise and Sports Science Australia (ESSA) and others. This ensures that the support offered is evidence-based, credible, and aligned with national and international public health recommendations. Content includes practical guidance on healthy eating, physical activity, cardiovascular health, digestive health, osteoporosis prevention, stress management, sleep hygiene, and the mental health benefits of exercise. The digital health coach provides guidance using structured conversational techniques based on participant data, aligned to health guidelines. For example: -After assessing vegetable intake, the coach offers tailored encouragement compared to the Australian Dietary Guidelines (e.g., highlighting strengths first, then suggesting simple improvements if vegetable serves are below recommendations). -Following assessment of physical activity levels, the coach compares current behaviour to national activity guidelines and offers achievable next steps (e.g., suggesting an additional walk if a participant is close to but below 150 minutes per week). -If a participant shares lower confidence or motivation, the coach adapts by suggesting smaller, habit-based goals rather than outcome-based goals (e.g., starting with 5–10 minutes of activity rather than full sessions). Goal-setting follows a structured process: Up to three goal areas are suggested based on participant priorities and opportunities for improvement, participant's select their preference or propose their own. Goals are personalised to the participant's lifestyle context, such as their available time, preferred activities, and readiness for change. Participant engagement is monitored through backend system APIs, with automated reporting of usage metrics. Participants with declining engagement patterns are sent supportive re-engagement emails encouraging continued involvement. Strategies to monitor adherence include: -Login success events (first and subsequent logins), -Page view tracking across the app, -Click tracking for key actions such as goal setup, health check completion, and dashboard/menu interactions, -Monitoring of review of check-in summaries, -Recording the frequency of check-ins (based on the participant-selected schedule), -Average session duration calculated from time spent within the platform.

Sponsors

University of South Australia
Lead SponsorUniversity

Study design

Allocation
Randomised controlled trial
Intervention model
Parallel
Primary purpose
Prevention
Masking
Blinded (masking used) (Investigator)

Eligibility

Sex/Gender
All
Age
18 Years to 74 Years
Healthy volunteers
Yes

Inclusion criteria

• BMI equal to or greater than 25 kg/m² • English speaking • Internet access • South Australian resident

Exclusion criteria

• Pregnancy/lactation • Clinical conditions requiring diet modification • Severe heart conditions e.g. unstable angina (heart conditions allowed with medical clearance) • Eating disorders • Severe food allergy (anaphylaxis) • Diabetes requiring medication • Use of weight loss drugs • End-stage organ disease • Epilepsy • Severe respiratory disease • Insufficient technology skills (must be able to send and receive text messages on a smart phone, navigate to a webpage, and upload a photo from a mobile device to a webpage)

Outcome results

None listed

Source: ANZCTR · Data processed: Feb 4, 2026