Wearanize+: A Multimodal Dataset for Evaluating Wearable Technologies in Sleep Research

• DOI: 10.31219/osf.io/dth8y_v1
• Publication Date: 2025-01-30T21:53:53Z
• Authors: Niloy Sikder; Lieuwe Verkaar; Anastasiya Paltar...; Selin Acan; Elena Krugliakova; Yevgenia Rosenblum; Matthias Krauledat; Martin Dresler; Paul Zerr
• Keywords: Computational Neuroscience; Neuroscience and Neurobiology; Life Sciences

Sleep research heavily relies on Polysomnography (PSG) recordings to assess sleep architecture. While effective, this method is time-consuming and requires substantial resources and labor. Modern wearable devices provide a promising alternative for sleep monitoring as they are easy to wear and maintain. However, these devices are constrained by a limited number of channels and comparatively lower data quality, which often leads to unreliable outcomes derived from partial readings. To address this, we propose using multiple wearable devices and combining their outputs to acquire reliable sleep data. However, the feasibility of this approach must be rigorously tested before being relied upon to supplant PSG in scientific studies. To facilitate this, we have curated a dataset with concurrent full PSG and wearable device recordings of overnight sleep sessions. This dataset, named Wearanize+, includes data from 130 participants (one night each) using three different wearable devices: Zmax headband, Empatica E4 wristband, and ActivPAL leg patch, alongside full-scale PSG recorded with SomnoScreen Plus and Mentalab Explore Pro. It also includes questionnaires, such as PSQI, MADRE, and PHQ-9, providing information on participants’ sleep, dreams, and overall health. This paper documents the setup, data collection, and data preprocessing steps of the Wearanize+ project, serving as a guide for using the resultant dataset. Our objectives with the dataset include developing machine learning models that can derive PSG-grade sleep stages from wearable devices’ data and exploring alternative data modalities for sleep scoring, especially when EEG signals are excessively noisy. Given its potential to support the testing of a wide range of hypotheses, we have made the dataset openly available to the research community.