As a multifactorial condition, investigations in the area of obesity can focus on a wide range of different types of behaviors, including those associated with physical activity, nutritional intake, sedentary behavior, and sometimes weighing behavior. Studies which require collecting accurate data on such behaviours will impose notoriously difficult and burdensome tasks to participants, which negatively affects their adherence.
Even when observed, it can be challenging to infer the motivation or intentions behind behaviours. For example, the reasons underlying a lack of physical activity can be a feeling of lack of social support, concerns regarding physical safety, an injury, or perception of adverse weather. We are also frequently interested in the role of exposures (to built environments, social environments, food environments), knowledge, attitudes and beliefs related to such behaviors, as well as dynamics of outcomes such as weight, BMI, motivation, and downstream factors such as injuries, including those of a subclinical nature. Such factors are again frequently difficult to measure accurately and with sufficient temporal resolution without overwhelming participants.
Given the multiple causal pathways involved, when interpreting outcomes in the context of interventions, it can be highly desirable understand which behaviours have been successfully altered (e.g., physical activity) and whether there are compensatory or simultaneous induced changes in other pathways (e.g., increases in caloric intake, or sedentary behavior).
Ethica can play a key role in easing such data collection, linking together knowledge of fine—grained exposures, dynamics across different particular pathways, and their time—varying effects on outcomes. By linking physical measures in the form of sensor data for factors such as physical activity, sedentary behaviour, exposures to environments using unobtrusive ecological momentary assessments (EMAs) and crowdsourcing, such understanding can span our knowledge on exposures, attitudes, and ideation. The potential for linking such information with physical measures recorded from smart watches and data from GIS, weather and other databases further emphasizes the potential for discovery.
Example Problems and Questions That Can be Addressed
To what degree has this lifestyle intervention budged sedentary behaviour, moderate—to—vigorous physical activity, or dietary intake? In intervened upon groups? In non—intervened upon individuals within the social networks of those intervened upon?
To what degree is changing physical activity compensated by elevated dietary intake?
To what degree does changing physical activity also lead to a decrease in sedentary behaviour?
Identifying reliable early indicators of an impending incident of dietary lapse or lower physical activity
What fraction of meals are people likely eating at home?
What exposures or ideations are likely associated with relapse?
What exposures or ideations are likely associated with adverse weight measurements?
How do changes in a given person’s target weight (if any) reflect weights of those around them?
To what degree are one individual's risk and protective factors likely to affect another person within their social network? (Physical activity, sedentary behaviour, diet, social connections)
How much and how soon does the intervention affect different pathways involving risk factors (e.g., diet, physical activity, sedentary behaviour, care seeking, etc.)
With dynamic modeling:
With a particle filtered dynamic model receiving a stream of accelerometer readings, weight measurements, possibly caloric consumption estimates, anticipating what lies ahead.
How would interventions to improve diet or physical activity on the part of one person be likely to affect their own trajectory, and that of their family and broader network?