Lea Nickel¹, Silke Schicktanz¹

¹ Institute for Ethics and History of Medicine, University Medical Centre Göttingen

Due to its image-based diagnostics and data-intensive treatment decisions, oncology is regarded as an important field of application for artificial intelligence (AI). AI tools are used in a variety of ways here: from automated image analysis and risk prediction to support in treatment decisions.

Rather than focusing on individual applications, in this talk we aim to map AI implementations along a ‘patient journey’ in order to highlight the multiplicity and interplay of many different AI applications, as well as their institutional embedding, and to facilitate critical discussion of these issues. This involves examining interaction with AI-supported medicine both from the perspective of a patient’s life story and from the perspective of a socio-technical impact assessment of such a future model of healthcare.

The ethical focus is on reflecting on patients’ agency: how can they distinguish between different AI systems in order to make informed decisions? At which points in the care pathway are opt-out options possible or even advisable? ‘Patient journey mapping’ can thus capture neglected biographical and procedural aspects of questions concerning patient autonomy, informed decision-making and epistemic uncertainty.

In addition, we discuss two challenges posed by this epistemic modelling approach: firstly, various visualisation strategies for such ‘journeys’ often convey implicit normative assumptions and narratives that can influence perception and ethical judgement. Critical reflection and the provision of argumentative justification for the chosen strategies are essential here. The second challenge is that AI systems currently differ in their stage of implementation, raising the question of how this can be methodologically addressed from the perspective of anticipatory applied ethics.

We argue that both retrospective experiences and prospective, visionary perspectives should be consistently incorporated into the ethical analysis and technical development of AI in oncology as patient needs. Despite remaining epistemic limitations, this approach opens up innovative perspectives for patient-centred AI integration.