Micro-EMA

Using “microinteractions” for longitudinal measurement of behavior

Duration: 

Jan 2016 – Present

Brief:

Ecological momentary assessment (EMA) is an in-situ data collection methodology (for digital phenotyping), where the user’s smartphone beeps/prompts several times a day (often 6 – 10 times) with a set of multiple-choice questions related to the research construct of interest. EMA is used by researchers to measure behaviors such as mood, eating habits, and current physical activity to name a few. However, EMA often induces high study burden on the participants resulting in lower compliance/response rates. Low response rates affect the measurement of behavior, thereby affecting the quality of behavioral data.

We implemented μEMA (micro-EMA), or microinteractions-based ecological momentary assessment. In μEMA, all the EMA beeps/prompts are reduced to single questions with Yes/No kind of answers – responding to each of these questions is a single-tap quick glanceable “microinteraction” (like checking the time on a wristwatch), taking hardly ~ 2s. μEMA leverages the quick access time and reliable tactile vibrations on the smartwatches to deliver short self-report surveys at a high temporal density (like a sensor).

Compliance assessment:

μEMA was evaluated in two pilot studies. First, we compared μEMA with traditional mobile-based EMA in a multiweek between-subjects study. Second, we compared μEMA with watch-EMA, which is mobile-based EMA directly translated on to a smartwatch. Our pilot studies show that μEMA offers:

  1. Significantly higher response rates than phone-based EMA
  2. Lower study burden on the participants, despite 8 times more interruption than traditional phone-based EMA
  3. High temporal density measurement of behavior (like an objective sensor) for long periods of time (e.g., 4-weeks or more). This is also demonstrated in a sample data collected using μEMA for physical activity measurement.

Finally, we also demonstrate that high compliance and temporal density of μEMA is due to the microinteractions on the smartwatch and not due to the use of smartwatch alone.

Ecological validation:

We are currently conducting studies to examine the ecological validity of μEMA surveys and hope to apply this data collection methodology in real-time activity recognition. Please watch this space as I update more results from the follow-up studies. The image below demonstrates how μEMA can be used to gather self-report data with a very high temporal density, without burdening the user.

This short video clip demonstrates how μEMA captures momentary changes in the behavior all using just quick glanceable self-report interactions.

Watch a detailed talk by Prof. Stephen Intille on μEMA:

Note: If you are a student or staff at Northeastern University owning an Android phone, you are eligible to participate in our studies with μEMA. Please get in touch with me to sign up for the study.

Technology Used: 

Android Wear, Firebase Web/Cloud Platform, Actigraph, Actilife, R, Python

Role: 

Interaction Design, Experiment Design, Statistical Analysis, Android Programming

Skills Developed: 

Actigraphy, Wearable Computing, Multilevel Statistics, Longitudinal Study Design

Related Publications:

Intille, S., Haynes, C., Maniar, D., Ponnada, A., Manjourides, J. (2016, September) μEMA: Microinteraction-based Ecological Momentary Assessment (EMA) Using a Smartwatch. In Proceedings of the 2016 ACM Joint Conference on Pervasive and Ubiquitous Computing (UbiComp 2016). 

Aditya Ponnada, Caitlin Haynes, Dharam Maniar, Justin Manjourides, and Stephen Intille. 2017. Microinteraction ecological momentary assessment response rates: Effect of microinteractions or the smartwatch? PACM Interact. Mob. Wearable Ubiquitous Technol. 1, 3, DOI: 10.11.45.

Collaborators: 

Binod Thapa-Chhetry (Ph.D. Student, Northeastern University), Krystal Huey (PharmD Student, Northeastern University), Prof. Justin Manjourides (Asst. Prof., Northeastern University)

Acknowledgment: 

This work was funded, in part, by a Google Glass Research Award. The phone-based EMA software system was made possible with funding from the NIH (R21 HL108018-01).