Most of us acknowledge the importance of a healthy lifestyle and are interested to pursue one. Yet, obesity rates keep increasing, and cardiovascular disease is the lead cause of death in Finland. For many, it is not the question of what needs to be done, but how. Succeeding in sustainable behavior change can be challenging amid the demands of everyday life. For example, busy schedules, work stress, financial problems, or mental health issues reduce our capacity to introduce changes to our lives.
Digital health behavior change interventions aim to guide and encourage people to learn and maintain healthy habits. Digital health interventions that are personalized to the needs and capabilities of a person are especially promising, as they hold the potential of delivering the right kind of support at the right time.
“Personalization could mean recommending behavior change goals that meet one’s needs, interests, and abilities. Or it could mean recommending daily actions that have been found useful among one’s peer group who share similar challenges and life situations. Increasing the difficulty of the suggested actions gradually as one’s skills develop is also an important personalization aspect that cultivates the personal feeling of success”, Honka explains.
The technical development of such personalized digital interventions requires maintaining a user model that keeps track of the data that describes the personal factors determining one’s behavior, the so-called user features. Personalization algorithms are used to select the appropriate guidance or feedback that match the user features.
In her dissertation, Honka describes the implementation of a prototype health recommender system that suggests behavior change actions based on users’ personal behavior change needs and readiness to take action. The system provided recommendations from various lifestyle and well-being domains, such as sleep, eating habits, physical activity, workload and stress management, value clarification, and alcohol consumption. The feasibility of the system was studied in the context of occupational stress-management coaching.
“The implemented health recommender system was able to produce suggestions that were appropriate and useful for the study participants. This indicates that an adequate personalization experience can be provided even with a rather limited user model. However, supporting a more extensive set of user features, such as users’ abilities and the possible barriers hindering behavior change, is expected to improve the suitability of the recommendations further”, Honka adds.
Recommender algorithms are popular in many commercial online services due to their scalability to large user and product bases that can be described with a multitude of features. Such characteristics are beneficial also for digital health interventions that are intended for wide use.
“Recommender algorithms offer promising techniques for personalizing digital health behavior change interventions. However, such interventions need to deal with sensitive user data. Therefore, it is crucial to ensure reliable data privacy and security practices before extensively personalized health interventions can be introduced to the public”, Honka reminds.
The dissertation also provides insights on how users’ personal values may be leveraged for personalization. Honka explored associations between the self-reported values, well-being, and health behaviors of Finns by analyzing a large web-survey dataset, which included more than 100,000 responses. People tended to report behaviors that were aligned with their values, which supports the prevailing view of values motivating behavior.
Indeed, knowledge of the values one holds important could be used to enhance commitment to personal health goals by associating health goals with value-based aspirations. For example, a career-focused person might become motivated to improve sleep and exercise habits once they realize how these habits influence productivity. The idea of utilizing values for the personalization of interventions is a new opening in the field of health technology.
Anita Honka conducted the research summarized in her dissertation while working as a Research Scientist at the Smart Health group at VTT Technical Research Centre of Finland. Nowadays, she is a Senior Data Scientist at Garmin. She lives in Vesilahti.
Public defense on Friday 8 March
The doctoral dissertation of MSc (Mathematics) Anita Honka in the field of health technology titled Personalization of digital health behavior change interventions for health promotion: A user modeling framework and a recommender system will be publicly examined at the Faculty of Medicine and Health technology at 12 o’clock on Friday 8 March 2024. The venue is City centre campus, Main building, auditorium D11. (Kalevantie 4, Tampere). The Opponent will be Emeritus Professor Anthony Maeder from Flinders University, Adelaide, Australia. The Custos will be Professor Mark van Gils from the Faculty of Medicine and Health Technology.