Abstract for: Operational feasibility and scaling dynamics of hospital-based production of personalised RNA therapies

Personalised RNA therapies offer promising treatments for rare diseases, yet many of the approximately 300 million people affected globally still lack effective treatment options. Recent advances make small-batch, on-demand production technically feasible, raising the possibility of producing personalised RNA therapies within hospital pharmacies. However, hospital pharmacies are complex socio-technical systems in which capacity, workforce, supply chains interact through feedback mechanisms. This study develops a system dynamics model to explore the feasibility and scaling dynamics of hospital-based production of personalised RNA therapies. A causal loop diagram (CLD) was developed using literature, expert interviews with hospital pharmacy leaders, and a participatory modelling workshop. The model represents how organisational RNA production capability emerges through interactions between demand, workforce, infrastructure, material capacity, and organisational learning. Preliminary analysis identifies reinforcing and balancing feedback structures shaping hospital production capability. Rising operational workload increases pressure to prioritise day-to-day production over capability development, slowing investments in workforce, infrastructure, and material capacity. Organisational learning can strengthen long-term production capability by improving efficiency and partially offsetting these constraints. The model hypothesises that the scalability of hospital-based RNA therapy production depends on balancing short-term operational demands with long-term investments in organisational capability. Delays in workforce development, infrastructure expansion, and material availability may constrain scaling, whereas sustained organisational learning strengthens production capability. Future work will translate the CLD into a stock-and-flow simulation model to evaluate implementation strategies. AI used only for language editing