Adoption of AI-enabled mental health wearables in India: The roles of psychological assurance and algorithmic credibility

This study examines the willingness to adopt AI-based wearable devices for mental health diagnosis, with a focus on the importance of psychological safety and algorithmic trust. It highlights the need for early and ongoing diagnosis of mental health issues, as AI-enabled wearable devices with advanced digital biomarkers and machine learning can help achieve this goal. Unlike studies on trust and TAM in digital health that focus on the perceived usefulness of the system, this study is the first to show that psychological safety is the dominant factor with stronger effects on adoption, illustrating a unique trust-emotion mechanism in AI-assisted wearables for mental health. Developing and validating a multidimensional adoption model, this study is grounded within the Technology Acceptance Model (TAM), extended with trust-based constructs and the Information Systems Success Model. Data were collected through a cross-sectional online survey of 763 respondents from urban and semi-urban areas in the Indian states of Odisha, Maharashtra, Delhi, and Tamil Nadu, representing a range of educational qualifications and occupations. Using SmartPLS 4, the findings suggest that psychological assurance is the dominant predictor of behavioral intention (β = 0.659, p < 0.001, 95% CI [0.608, 0.709]). While there was no direct effect on behavioral intention (β = −0.023, p = 0.452), a significant pathway trust was observed. Algorithmic credibility positively influenced psychological assurance (β = 0.276, p < 0.001) and perceived diagnostic accuracy (β = 0.555, p < 0.001), suggesting an indirect effect. Psychological assurance was positively influenced by perceived usefulness (β = 0.301, p < 0.001), and a positive effect was observed on behavioural intention (β = 0.062, p = 0.029) in relation to perceived diagnostic accuracy. Finally, the authors concluded that a range of influencing factors affects perceived diagnostic accuracy, with a moderate effect (β = 0.102, p = 0.003) also being found. These results underscore the importance of examining both the emotional trust aspect and the technical accuracy aspect of AI-enabled mental health wearables. For practitioners and policymakers, the findings underscore the importance of focusing on explainable AI, clinician endorsement, and reassurance feedback loops to enhance the adoption of mental health technologies and improve mental health care outcomes.