Human-AI Collaboration for Goal Achievement in Physical Activity Tracking

Physical activity (PA) is fundamental to long-term health, yet many people struggle to sustain engagement with existing PA tracking systems. Despite their widespread adoption, current trackers offer limited support for personalised, adaptive guidance throughout users’ physical activity journeys which leads to disengagement, disrupted behavioural routines, and reduced potential for long-term behaviour change. While prior work has incorporated AI to generate personalised predictions and recommendations, such systems typically treat both objective and self-reported data as inputs for modelling rather than as resources for ongoing negotiation, and focus on isolated functions such as goal prediction instead of supporting users across the entire PA journey. This PhD aims to redefine how AI can contribute to behaviour change by investigating human–AI collaboration across the entire PA cycle, including goal setting, planning, tracking, and reflection. Instead of positioning AI as a prescriptive analytics engine, this work explores how AI can act as a collaborative partner that adapts to users’ behavioural context, fosters user’s agency, self- efficacy, and strengthens long-term engagement. The research will generate new interaction mechanisms, transparency models, and human-AI collaborative PA tracking strategies. Ultimately, this work seeks to advance foundational understanding of effective human–AI collaboration for behaviour change and provide design knowledge for the next generation of personalised, trustworthy PA support systems.