Digital phenotyping and digital monitoring technologies for relapse detection in mental health: a systematic review

Mental health relapse remains a major challenge in the long-term management of psychiatric disorders. Conventional monitoring approaches rely primarily on periodic clinical assessments and self-report measures, which may fail to capture early behavioural changes preceding symptom deterioration. Digital phenotyping, which refers to the continuous collection and analysis of behavioural and physiological data from personal digital devices has emerged as a promising approach for monitoring mental health trajectories and identifying early warning signals of relapse. However, the evidence base remains fragmented, with significant variability in methodologies, populations, and outcome measures, limiting clear conclusions. This systematic review synthesises the current evidence on digital phenotyping and related digital monitoring approaches used to detect, predict, or prevent relapse in individuals living with mental health conditions. The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) guidelines and was prospectively registered in PROSPERO (CRD42024561513). A comprehensive literature search was conducted on 17 December 2024 across multiple databases, namely PubMed, Scopus, Web of Science, IEEE Xplore, CINAHL, ACM Digital Library, and Google Scholar. Eligible studies should have examined digital phenotyping or digital monitoring technologies in the context of relapse detection, prediction, or prevention in mental health conditions. Both experimental and observational study designs were included, encompassing randomised and cluster trials, along with pilot, feasibility, and observational monitoring studies. Data were extracted on study characteristics, digital technologies, monitoring modalities, and relapse-related outcomes. Risk of bias for randomised controlled trials was assessed using the Cochrane Risk of Bias (RoB) 2 tool. Twenty-two studies involving approximately 12,000 participants met the inclusion criteria. Most studies were conducted in high-income countries and evaluated a diverse range of digital monitoring technologies, including smartphone applications, wearable sensing devices, SMS-based interventions, and predictive algorithms. Across studies, digital monitoring technologies demonstrated the potential to identify behavioural signals associated with worsening mental health symptoms and early relapse risk, particularly through smartphone-based monitoring and digital therapeutic platforms. However, the evidence base remains heterogeneous, with many studies focused on feasibility or pilot evaluations rather than validated relapse prediction systems. SMS-based interventions and app-based cognitive behavioural therapy tools generally reported positive findings, whereas wearable devices and predictive algorithm approaches showed mixed findings. Digital phenotyping and related digital monitoring approaches show promise for improving relapse monitoring and early detection in mental health care. However, the current evidence base remains heterogeneous. Future research should prioritise larger longitudinal studies, standardised relapse definitions, and multimodal monitoring approaches integrating behavioural, physiological, and clinical data to improve the reliability and clinical applicability of digital phenotyping technologies. CRD42024561513. Not applicable.