This study presents a multi-physics computational framework to simulate and optimise the performance of subcutaneous insulin port delivery systems used in diabetes treatment. The approach integrates syringe injection dynamics, pharmacokinetics and tissue permeability modelling to comprehensively represent the key processes involved. In particular, the phenomenon of pressure-related damage of the soft tissue consequent to an injection has been studied. This disrupts the normal organization of adipocytes and can lead to lipodystrophies, which inhibit insulin diffusion and absorption. By combining a real-case geometry model with a weakly-coupled physical model, accounting for both fluid distribution into the tissue and absorption into the blood stream, the system response has been analysed. Healthy and compromised tissue initial conditions have been considered, to study opposite behaviours and to assess the impact of pre-existing tissue defects. Within the computational strategy, a particular emphasis has been placed on the modelling of soft tissue damage, linking local pressure distribution to alterations of permeability and accumulation effects induced by recurrent injections. With the proposed framework the efficiency of the device has been optimised, redistributing potential damage across a wider volume and ultimately mitigating the adverse effects of delayed insulin absorption.