Between Surveillance and Support: A Qualitative Study of Tuberculosis Patients' Expectations and Concerns About...

This study explores how tuberculosis (TB) patients in China perceive AI-assisted remote health services, focusing on the psychological and sociocultural dynamics involved in balancing perceived support and perceived surveillance.

A qualitative descriptive approach was adopted. 25 TB patients were recruited from urban and rural health facilities in Hubei Province, including both those currently in treatment and those who had recently completed it. In-depth, semi-structured interviews were conducted to examine patients' treatment experiences, digital literacy, and attitudes toward AI-assisted care. The AI system described to participants was a hypothetical prototype based on emerging technologies rather than an implemented service. Thematic analysis was guided by the Health Belief Model and Affordance Theory to identify key patterns and interpret their meanings.

Five key themes emerged. Patients reported treatment fatigue and fluctuating motivation, reflecting complex psychological demands. Trust in AI systems was conditional, shaped by concerns about usability, digital unfamiliarity, and system reliability. Participants experienced a tension between viewing AI tools as supportive and feeling uncomfortable with constant monitoring, especially given the stigmatized and regulated nature of TB. A strong desire to preserve autonomy and dignity shaped patients' preferences for systems that minimize disruption and allow self-regulation. Acceptability was influenced by interface simplicity, preferred modalities such as voice-based prompts, and the assurance that AI would supplement rather than replace human care. These findings were synthesized into a conceptual framework, illustrating how treatment burden, psychological interpretations of AI, and perceived empowerment converge into a process of contextualized acceptance.

This study offers new insight into digital health engagement among an underserved population. It shows that TB patients do not passively receive AI interventions but interpret and evaluate them in light of their experiences and expectations. Designing acceptable AI-assisted systems requires sensitivity to patients' social contexts, emotional needs, and desire for agency in care.