Ecological momentary assessment (EMA) is a data collection method that assesses individual's experiences as they occur in real time and in the natural environment. Its usefulness has been limited, however, by the available technology and the burden it places on the participant. We propose to improve EMA by further developing and testing an adaptive assessment system designed by our interdisciplinary team using state-of-the-science hardware and software that limits subject burden while facilitating real-time data collection. Our modifications to EMA are designed to increase data quality by optimizing the sampling schedule for random prompts measuring momentary covariates, while reducing the burden on study participants. We hypothesize that our proposed method will mitigate limitations associated with EMA and provide a means to quantify variables not traditionally measured in EMA, e.g., duration and quality of sleep, physical activity, daily weight and location, in a study of intentional, supervised weight loss in adults. We will test our hypothesis by using real-time transmission of data to link information from smart-phones, weight scales, daily diaries, actigraphs, and accelerometers. Modified-EMA sampling will be response adaptive - increasing the frequency of random assessments in response to indicators of increased risk, such as low mood for 3 days or high levels of stress. The value of the models we propose to further develop here will be applicable to a wide range of conditions in which the process of self-imposed behavior change is maintained or reversed, including adherence to self-management of chronic diseases (e.g., diabetes, kidney disease, cancer). Additionally, the framework for EMA sampling design and model fitting that we propose to develop is anticipated to be broadly applicable not only to EMA, but also to survival analysis in biomedicine, spatial epidemiology in environmental health, and to event history data in the social and behavioral sciences.
We suspect that individuals' moods, feelings and environments affect their behaviors, and the best way to understand their effect on behaviors is to collect data throughout the day, in real- time with participant in their real environment using a smart phone so data can be transmitted in real time. However, there is a balance between collecting data too frequently and potentially over-burdening participants, and collecting data too infrequently and not being able to answer our research questions. Thus, we will examine the data to determine how frequent and the best times of day we should ask participants to answer our brief survey questions via the smart phone. For our study we are researching relapse following intentional weight loss, but more importantly, we are advancing a data collection method that can be applied to a wide variety of health behaviors.
