English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Personal sensing, leveraging data passively and near-continuously collectedwith wearables from patients in their ecological environment, is a promisingparadigm to monitor mood disorders (MDs), a major determinant of worldwidedisease burden. However, collecting and annotating wearable data is veryresource-intensive. Studies of this kind can thus typically afford to recruitonly a couple dozens of patients. This constitutes one of the major obstaclesto applying modern supervised machine learning techniques to MDs detection. Inthis paper, we overcome this data bottleneck and advance the detection of MDsacute episode vs stable state from wearables data on the back of recentadvances in self-supervised learning (SSL). This leverages unlabelled data tolearn representations during pre-training, subsequently exploited for asupervised task. First, we collected open-access datasets recording with anEmpatica E4 spanning different, unrelated to MD monitoring, personal sensingtasks -- from emotion recognition in Super Mario players to stress detection inundergraduates -- and devised a pre-processing pipeline performing on-/off-bodydetection, sleep-wake detection, segmentation, and (optionally) featureextraction. With 161 E4-recorded subjects, we introduce E4SelfLearning, thelargest to date open access collection, and its pre-processing pipeline.Second, we show that SSL confidently outperforms fully-supervised pipelinesusing either our novel E4-tailored Transformer architecture (E4mer) orclassical baseline XGBoost: 81.23% against 75.35% (E4mer) and 72.02% (XGBoost)correctly classified recording segments from 64 (half acute, half stable)patients. Lastly, we illustrate that SSL performance is strongly associatedwith the specific surrogate task employed for pre-training as well as withunlabelled data availability.

Wearable data from subjects playing Super Mario, sitting university  exams, or performing physical exercise help detect acute mood episodes via  self-supervised learning