A Computerized Approach to Tailor Health Messages for Enhancing Physical Activity in People with COPD: System Design, Development and Feasibility study

Engaging in physical activity (PA) is essential for individuals especially those with COPD to uphold their quality of life and prevent exacerbations. Despite its significance, there is a widespread issue of low PA levels and a high dropout rate in PA interventions among COPD patients. Computer tailored health communication (CTHC) has demonstrated greater effectiveness in promoting behaviour change compared to generic health information or no information across various populations. CTHC employs algorithms to automatically select personalised health information, making it scalable to reach a large audience.

Behaviour change theories provide the foundational principles for constructing CTHC systems. While CTHC has been successful in enhancing PA levels in individuals with long-term conditions, there is limited evidence for its effectiveness specifically in promoting PA for people with COPD (PwCOPD). Previous studies lack information on the interactive effects between tailored health information and the psychological impact on individuals. This knowledge gap hinders understanding how chosen tailoring factors influence targeted behaviours and poses challenges in adapting CTHC for diverse populations and new behaviours.

To address this gap and to identify effective tailoring factors for CTHC in promoting PA for COPD patients, this project investigates the feasibility of employing CTHC methods that combine rule-based and machine learning based classifiers. The aim is to provide tailored health messages and assessing the algorithms’ performance through participant ratings, which indicate the psychological impact of the messages on the participants. The proposed system underwent testing in both healthy population and Chinese PwCOPD. The rule-based classifier proves useful when historical user data is limited, while machine learning classifiers significantly enhance system performance with ample data.

The results offer insights into the relationships among various tailoring factors, algorithms,
and user attitudes toward selected messages. These findings suggest that building a successful CTHC system is feasible, capable of delivering tailored health messages that induce cognitive changes and subsequent behaviour changes to promote PA for individuals with COPD in China.