A Machine Learning Approach to Identifying Healthy Aging: Developing Healthy Aging Phenotypes and Personas in the Canadian Longitudinal Study on Aging

Demographic forecasts indicate that Canada's population structure will change dramatically over the next two decades, such that 25% of the population will be over 65 by 2036. Optimizing population health and wellness over the trajectory of aging - i.e. optimizing 'healthy aging' - is therefore a major research and policy goal in Canada. Researchers often explore what it means to age "successfully" or in a "healthy way" by applying a biomedical conceptual framework, and often the general public's perspective is not taken into consideration. However, there is increasing recognition that a lay perspective of healthy aging is needed. Our team has recently used new techniques such as natural language processing to categorize open-text data on lay perspectives of healthy aging in the Canadian Longitudinal Study of Aging (CLSA), identifying ten themes in responses to the interview question, 'What do you think makes people live long and keep well?' Now, our aim is to look at the characteristics of people, or 'phenotypes' who are in each of the 10 themes. These phenotypes will be based on social, psychological, behavioural, physiological, and functional information collected in the CLSA. Further, we will look to see if there are clusters of individuals based on these phenotypes, which we will use to generate 'personas'. We will examine how they relate to self-rated assessment of healthy aging based on the question, 'In terms of your own healthy aging, would you say it is excellent, very good, good, fair, or poor?' and how they relate back to the themes generated based on perceptions of what healthy aging means. Examining perceptions of healthy aging, self-rated assessments of healthy aging and individual characteristics and how they are inter-related will help us to understand lay perspectives on what it means to age in a healthy way.