<aside> <img src="/icons/reorder_gray.svg" alt="/icons/reorder_gray.svg" width="40px" /> Home

Team Members

User Survey

FOMO

🔜 Coming Soon

</aside>

Study Objective: Identify pain points in data processing and analysis among researchers in our lab using commercial wearable devices.

Subjects: PhD students currently engaged or with previous experience in commercial wearable device research in the lab (n=5).

Challenges of Data Management in Wearable Device Studies

Theme 1: Establishing and Justifying Data Missingness Criteria in Longitudinal Studies

Theme 2: Laborious Data Preprocessing and Alignment Across Multiple Data Streams

**Theme 3:**Variability and Dependence of Device-Specific Metrics on Proprietary Algorithms

“And I think we spent two weeks just figuring out the missingness metrics we're going to be using. We also got a little pushback from the reviewers because they simply asked us to expand on why we used the missingness criteria that we've used because there's not enough literature on it.”

“I think the most [time-consuming] is going to be the quality check, which involves basically which participants to involve in my study and who to drop. So, for example, drawing out a consort diagram, making sure who needs to be dropped, and getting those numbers, basically.”\

Takeaway

• Developing missingness criteria is challenging due to the absence of standardized guidelines and the diversity of study requirements.
• Preprocessing and aligning data across multiple streams is labor-intensive and demands significant manual effort.
• Routine data quality checks, such as calculating wear time, can be unexpectedly time-consuming and frustrating.
• Dependence on device-specific metrics and algorithms necessitates customized data handling strategies.

Finding and Utilizing Existing Resources and Knowledge