IEEE-EMBS BHI 2025

Tutorial: Patterns of Missingness in Wearables Data

October 29, 2025 (8AM) | 📍 Georgia Tech Global Learning Center (Atlanta, GA)

Join us to gain practical tools, real-world insights, and powerful frameworks that will transform missing data from a challenge into an opportunity for more robust discoveries! (Learn more)

Hero - Elements Webflow Library - BRIX Templates

Tutorial Materials

Download App - Elements Webflow Library - BRIX Templates

Slides

The tutorial slides can be accessed below. If you identify any errors or have recommendations for supplementary materials, we'd love to hear from you!

Create Account - Elements Webflow Library - BRIX Templates

Transform a missing data vector from a longitudinal Bring-Your-Own-Device study into cyclical representations of missingness!

We are looking for ways to expand our Python package: domains (EHR, wearables, etc), imputation methods, inference. Please fill out our  contribution form with relevant information.

Schedule

We want to understand how researchers from different backgrounds approach missing data. Your responses are anonymous and will help us evaluate learning outcomes, identify common challenges, and improve future tutorials.
Background
[slides]
What are digital biomarkers?
Nature of Missingness: What are the key types and mechanisms of missing data in wearables, and how do they affect longitudinal research?
Gaps in Literature
[slides]
How is missingness currently addressed in the literature, what gaps remain, and why do they matter for our research?
Case Studies
[slides]
Case study #1: Signal in the silence: What wear time reveals about postpartum depression [paper1, paper2]

Case study #2: Defining the Habitome: Phenotypes of Routine and Their Relationship to Health Outcomes [paper]
Hands on Tutorial
[Google Colab]
How can we practically detect, characterize, and utilize missingness in wearable datasets to strengthen analysis?

Acknowledgement:  We gratefully acknowledge All of Us participants for their contributions, without whom this research would not have been possible. We also thank the National Institutes of Health’s All of Us Research Program for making available the participant data examined in this study. This tutorial was funded by NSF CAREER Grant #2339669.

Presenters

Assistant Professor
Duke University

PhD Student
Duke University

PhD Student
Duke University

Undergraduate Student
Duke University

Postdoctoral Researcher
UNC-Chapel Hill