Designing and Testing a Physical Activity App with a Chatgpt-Based Chatbot

Background and objective: ChatGPT shows potential as a tool for creating content for health promotion. For example, personalised physical activity (PA) intervention messages can be generated efficiently using ChatGPT. We aimed to develop and test a chatbot to promote PA. Methods: We developed FysBot, a PA app with a chatbot powered by ChatGPT. Personas were created to prime user queries before prompting the chatbot. We conducted Think-aloud sessions to evaluate the usability of FysBot, including the chatbot. Results: The five adults who participated in the usability testing valued the PA recommendations and instructions from the chatbot but noted issues with its responses, proposed hiking trips, and step count accuracy, along with suggestions for improvements. Conclusion: Study participants indicated satisfaction with the chatbot's potential to motivate users to increase their physical activity. Addressing usability concerns related to user interaction and making the chatbot more personalised will enhance broader adoption in the future.