Designing for Centralized Healthcare: A Socio-Technical Case Study of Pancreatic Cancer Diagnosis and AI Integration

Pancreatic cancer remains one of the deadliest malignancies, with outcomes dependent on timely diagnosis and accurate assessment of surgical resectability. Centralized care models aim to improve diagnostic quality by concentrating specialized expertise. While centralization improves access to expertise, it introduces new challenges in coordinating diagnostic work across institutions, aligning heterogeneous data and technologies, and sustaining reliable collective decision-making under time pressure. How diagnostic reasoning is practically accomplished within these distributed care networks remains insufficiently understood. We conducted a qualitative case study of a regional centralized pancreatic cancer care network in the Netherlands. Drawing on socio-technical systems theory, we analyzed diagnostic workflows including imaging assessments, multidisciplinary team meetings, and surgical procedures through contextual observations. Our findings show that diagnostic reliability emerges from the coordinated interaction of distributed expertise, clinical data, and technological infrastructures. Multidisciplinary meetings, shared imaging systems, and iterative data interpretation play a central role, while tensions arise around efficiency, standardization, authority, and uncertainty. This study contributes (1) empirical insight into the socio-technical dynamics of centralized pancreatic cancer care, (2) a qualitative approach for studying complex, distributed healthcare systems, and (3) practice-grounded implications for integrating AI in ways that support existing clinical coordination and expertise.