Wearable Artificial Intelligence for Anxiety and Depression: Scoping Review

Mental health disorders represent a significant global burden, yet access to specialized psychiatric care remains severely limited in many regions. This scarcity of professional resources has created a pressing need for scalable, accessible diagnostic solutions. Wearable technology offers a potential pathway to bridge this gap by providing continuous physiological monitoring. Prior research has shown that integrating machine learning into portable sensors might facilitate remote mental health tracking. That uncertainty drove the need to evaluate how these systems are currently deployed in clinical or research settings. No prior work had resolved the full scope of existing applications for these specific mental health conditions. This gap motivated a comprehensive examination of the current landscape of digital health tools. The following synthesis clarifies how these devices are utilized to support individuals experiencing psychological distress.

The aim of this review is to explore the features of digital tools used for anxiety and depression. Researchers sought to identify primary application areas and highlight existing research gaps. This investigation addresses the challenge posed by the global shortage of mental health professionals. By evaluating current literature, the team clarifies how these technologies provide mental health services. The study specifically targets the functionality of systems that integrate machine learning with portable sensors. This work provides a foundation for understanding the current capabilities of these diagnostic platforms. The authors intended to map the landscape of existing research to guide future technological development. This systematic analysis serves to inform both clinicians and developers about the current state of the field.

Review Approach involved a systematic search across eight electronic databases to identify relevant literature. The team screened records from MEDLINE, PsycINFO, Embase, CINAHL, IEEE Xplore, ACM Digital Library, Scopus, and Google Scholar. Two independent reviewers performed the study selection and data extraction processes to ensure consistency. The investigators also examined citations from the included papers to capture additional pertinent research. Narrative synthesis served as the primary strategy for aggregating the extracted information. The researchers focused on identifying application areas and open research issues within the field. This methodology allowed for a comprehensive overview of the current state of digital health diagnostics. The study design prioritized transparency and rigor in summarizing the diverse findings across the selected publications.

Key Findings From the Literature reveal that 69 studies met the inclusion criteria out of 1203 initially identified records. Approximately two-thirds of these investigations focused on depression, while the remaining portion addressed anxiety. The most frequent application of these systems involves the diagnosis of mental health conditions. Researchers observed that none of the analyzed studies utilized these devices for therapeutic treatment purposes. Most participants in the included research fell within the 18 to 65 age range. Wrist-worn hardware represented the most common form factor for these monitoring devices. Physical activity data constituted the most frequently used category for training machine learning models. Random forest algorithms appeared as the most common computational approach for analyzing the gathered information.

Synthesis and Implications suggest that these digital tools hold significant promise for enhancing mental health service delivery. The authors propose that these systems are currently well-suited for the prescreening assessment of psychological conditions. Future investigations should prioritize statistical meta-analyses to rigorously evaluate the performance metrics of these diagnostic models. The researchers suggest that technology developers should increase investment in creating therapeutic features for these platforms. Currently, the literature indicates a complete absence of studies utilizing these devices for active clinical treatment. The authors highlight that the field remains in an early stage of development regarding practical intervention. These findings imply that while diagnostic capabilities are advancing, the transition to clinical management requires further innovation. The team emphasizes that broader adoption depends on verifying the effectiveness of these algorithms in diverse populations.