On 8 July 2026, Dr David Lyreskog presented the poster "Parkinson's Disease Management in the Era of AI: Navigating the Living Systems of Trust and Care" at the Future of Psychiatry event, organised by the Department of Psychiatry, University of Oxford.

The paper argues that current approaches to "Trustworthy AI" – based on fixed compliance principles such as fairness, transparency, and robustness – are insufficient for clinical contexts involving longitudinal care, distributed responsibility, and evolving patient-clinician relationships.

Using AI-PROGNOSIS as a case study, it proposes reconceptualising trustworthiness as a property of dynamic socio-technical systems, built through three principles:

• Contextual Integrity (calibrating data governance to clinical context, not generic benchmarks),

• Dynamic Responsibility (responsibility that shifts as AI models evolve), and

• Participatory Reflexivity (continual involvement of patients, carers, clinicians, and developers in governance).

The paper concludes that as AI becomes an active participant in psychiatric care, the central challenge is designing systems that remain trustworthy as they evolve – not simply building more intelligent algorithms.