Digital Biomarkers of Stress and Anxiety: Assessing the Quality and Diversity of Wearable Sensor Datasets

Read full text Summarise with VIRAT
Authors: Georgia Kalitsi
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Kalitsi G, Digital Biomarkers of Stress and Anxiety: Assessing the Quality and Diversity of Wearable Sensor Datasets, MSc Thesis, Master of Science in Biomedical Informatics, School of Medicine, Democritus University of Thrace, and ATHENA Research Center, Alexandroupoli, Greece, March 2026. Supervisors: E. kaldoudi, S. Didaskalou, A.Fiska This thesis explores the rapidly evolving field of digital biomarkers, focusing on the use of wearable sensors for the detection and monitoring of stress and anxiety. As the global burden of mental health conditions increases, wearable technology offers a transformative opportunity for the continuous, passive, and objective monitoring of physiological states in real-world settings. However, despite the proliferation of research in this domain, the field remains characterized by extreme heterogeneity. Datasets generated by various research groups often utilize different hardware, sampling frequencies, and stress-induction methodologies, creating a harmonization gap that prevents the development of generalizable, sensor-agnostic artificial intelligence models. The theoretical foundation of this work is built upon the evolution of stress research, moving from traditional clinical snapshots and biochemical markers toward modern autonomic nervous system monitoring. The research identifies the core signals necessary for robust digital biomarkers via a systematic search, following the PRISMA 2020 statement, targeting peer-reviewed studies published over the last decade. This search, performed via PubMed, identified 63 unique datasets which were then meticulously mapped based on their technical specifications, cohort demographics, and accessibility. Results indicate that the field of digital stress monitoring has transitioned from exploratory studies to an era of intensive data collection driven by the maturation of wearable technology, yet it remains hindered by significant data accessibility barriers and a persistent reliance on controlled laboratory environments over real-world settings. While physiological sensing is anchored by modalities like Electrodermal Activity (EDA), heart rate, and accelerometry, there is a clear shift toward higher-dimensional datasets and a strategic preference for more robust, motion-resistant signals in "in-the-wild" applications to ensure ecological validity. Despite the standardization of laboratory stressors and the frequent use of self-reported metrics for annotation, the extreme heterogeneity in hardware and methodologies has created a harmonization gap that prevents the fusion of these disparate data sources. To address this, the thesis proposes a 7-Tier Harmonization Roadmap as a structured, hierarchical framework to integrate heterogeneous data and bridge the gap between technical interoperability and physiological compatibility across diverse sensing ecosystems. While the study is limited by its primary focus on PubMed-indexed literature and the inherent inconsistencies in "ground truth" labeling across different studies, it provides a foundational strategy for the scientific community. The work signifies a shift from sensor-specific silos toward a unified data ecosystem. Ultimately, this thesis argues that the future of reliable digital stress biomarkers depends not on the creation of more isolated datasets, but on the systematic harmonization of existing resources to build truly generalizable and clinically relevant diagnostic tools.

Fields of Study & Subjects

Emotion and Mood Recognition Digital Mental Health Interventions Mental Health Research Topics

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
13 March 2026
Language
EN
Document type
Thesis