DIGITAL PHENOTYPING AS A NOVEL CLINICAL BIOMARKER: ASSESSING THE DIAGNOSTIC EFFICACY, ARTIFICIAL INTELLIGENCE INTEGRATION, AND SOCIO-ETHICAL IMPLICATIONS IN PSYCHIATRIC CARE

Abstract

Modern psychiatry relies heavily on subjective clinical assessments and patient reports, which often limits diagnostic accuracy. Digital phenotyping, defined as the collection of objective, continuous behavioural data using smartphones and wearables, is emerging as a promising source of novel clinical biomarkers. This review article aims to critically assess the diagnostic efficacy of digital phenotyping, its integration with artificial intelligence (AI) algorithms, and the socio-ethical implications in psychiatric care. An analysis of the available literature indicates that data derived from passive and active monitoring can be successfully used to objectify diagnosis and predict symptom exacerbations, including in affective and psychotic disorders. The integration of digital biomarkers with machine learning models enables the identification of hidden behavioural patterns, paving the way for personalised medicine. Despite their enormous clinical potential, the widespread implementation of these technologies faces significant barriers. These include socio-ethical challenges, such as privacy protection, data security, difficulties in obtaining fully informed consent, as well as the risk of stigmatisation and algorithmic bias. Furthermore, the continuous monitoring of patients raises questions about the impact of technology on trust and the therapeutic relationship. In summary, although AI-supported digital phenotyping represents a potential breakthrough in psychiatry, its safe implementation requires rigorous clinical validation and the establishment of a robust ethical framework to protect patients.