Artificial Intelligence Conversational Intervention for Encouraging Physical Activity in Older Adults

Artificial Intelligence Conversational Intervention for Encouraging Physical Activity in Older Adults Sadiq Sani, Kay Cooper, Nirmalie Wiratunga, Stewart Massie (Robert Gordon University); Ehud Reiter (University of Aberdeen); David Sim (Openbrolly)

Digital health interventions are often delivered in the form of notifications on a mobile phone. Despite the popularity of this approach, there is little evidence to indicate that mobile notifications are effective at promoting positive behaviour change, particularly in the long-term. The main problem is that text notifications offer one-way communication (from the device to the user) and hence provide no opportunity for interaction. In addition, mobile notifications are easily ignored; fewer than 30% of received notifications are typically viewed by users with average delays of close to 3 hours, highlighting the need for an alternative approach. Futhermore, the use of mobile text alerts can be problematic for older adults, who may find it difficult to read from a small screen and to interact with small menus and interfaces. So could conversational feedback and input make health technologies more accessible to older adults?

The aim of this study is to produce a conversational system that is able to monitor the physical activity levels of older adults from sensor data and provide a personalised conversational intervention to increase their physical activity. Personalised interventions are known to be more effective than generic health behaviour change interventions, so the conversational interventions will be tailored to individual’s needs and preferences. For example, identified individual motivators for being physically active could be used by the conversational system to encourage physical activity when inactivity is detected. The conversational intervention will be co-created by the research team and older adults, who will drive the development of feedback and methods of interaction.
Research outputs will include:

  1. An understanding of the feasibility and acceptability of conversational intervention for promoting physical activity;
  2. A set of design specifications for the conversational intervention system, co-created with the user group. These specifications will provide detailed use cases of how users will interact with the system, conditions under which dialogue interventions are deemed feasible and when they are not, types of dialogue the system is expected to support (e.g. questionnaire response gathering, information provider, reflection initiation) and type of information the system is expected to provide (e.g. education, performance analytics, suggestions)
  3. Specifications for how such a system can be developed using state of the art algorithms and technologies: a feasibility report detailing current algorithms and existing technologies that are available to support development of such a system. This report will present detailed evaluations of alternative solutions with the view to highlighting strengths, limitations and cost implications of each, in order to inform the choices that need to be made for further development of the conversational system.
  4. Data that can be used to inform a grant application to conduct a large-scale study to evaluate the effectiveness of the conversational intervention.

Our vision is to have a natural, ubiquitous and proactive system that can use conversation to deliver expert guidance and personalised advice to users.