Digital phenotyping via smartphones and wearables for assessing chronic stress, self-regulation, and interoception is in early stages, with machine learning accuracy ranging from 56.8% to 79%.
Digital phenotyping using wearables and smartphones is a promising but early-stage approach for assessing chronic stress, self-regulation, and interoception in adults.
BACKGROUND Digital phenotyping, the real-time quantification of human phenotype in situ via digital devices, offers new opportunities to understand how behavior change interventions influence brain health. This approach has gained increasing interest in recent years, especially in brain health research. Interoception, chronic stress, and self-regulation are key domains, and benefit from real- world, continuous assessment beyond what traditional methods can provide. OBJECTIVE The aim of this scoping review was to map and synthesize the literature of the last five years on the use of digital phenotyping to measure or predict interoception, chronic stress, and self-regulation in adults. The review addressed specific questions concerning the types of devices and sensors used, the psychological domains targeted, the nature of the data collected, feature extraction and intended targets, data processing methods, and the technological platforms utilized. METHODS We conducted a scoping review following the Joanna Briggs Institute (JBI) methodology and the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) guidelines. Eligibility criteria included studies using digital phenotyping to assess or predict interoception, chronic stress, or self-regulation in adults aged 21–65 years, based on data collected via smartphones or commercial wearable devices. Studies published since 2018 were considered. A systematic search of PubMed, Web of Science, and Scopus retrieved 850 records; one additional study was identified through Google Scholar. Two reviewers independently screened titles, abstracts, and full texts. Eighteen studies met the inclusion criteria. Results were synthesized narratively. RESULTS Of the 18 studies included, 11 addressed chronic stress or stress reactivity, 5 self-regulation, and 2 interoception. To gather physiological and behavioral data, 13 used wearable devices, 3 smartphones, and 2 both; 8 used smartphones for ecological momentary assessment (EMA). Heart rate variability (HRV) was the most common physiological measure (n=14), followed by electrodermal activity (EDA) and heart rate (HR) (n=4 each). Behavioral data (e.g., smartphone use, sleep, activity) was analyzed in 9 studies. Six studies used machine learning models, but only 3 reported classification accuracy (56.8–79%). Eight applied statistical methods to link features with stress or interoception, while 4 assessed self-regulation using predefined features without identifying new biomarkers. CONCLUSIONS This scoping review highlights that research on digital phenotyping for interoception, chronic stress, and self-regulation is still in its early stages, with most studies focusing on chronic stress and relying primarily on wearable devices. Integration of smartphone sensing and long-term monitoring remains limited, and reported analytical performance is generally modest. Despite these challenges, the widespread use of smartphones and wearables positions digital phenotyping as a promising and scalable approach for assessing brain health in daily life. Future research should emphasize longer-term, multimodal data collection, innovative analytical methods, and transparent reporting to advance the field. CLINICALTRIAL
Alvarez-Ambrosio et al. (Mon,) conducted a review in Interoception, chronic stress, and self-regulation (n=18). Digital phenotyping (smartphones and wearable devices) was evaluated on Use of digital phenotyping to measure or predict interoception, chronic stress, and self-regulation. Digital phenotyping via smartphones and wearables for assessing chronic stress, self-regulation, and interoception is in early stages, with machine learning accuracy ranging from 56.8% to 79%.