We investigate the application of semantic information theory to drug delivery systems (DDS) within the molecular communication (MC) framework. To operationalise this, we observe a DDS as a molecular concentration-based channel. Semantic information is defined as the amount of information required for a DDS to achieve its therapeutic goal in a dynamic environment. We derive it by introducing interventions, defined as modifications to DDS parameters, a viability function, and system-environment correlations quantified via the channel capacity. Here, the viability function represents DDS performance based on a drug dose-response relationship. Our model considers a system capable of inducing functional changes in a receiver cancer cell, where exceeding critical DDS parameter values can significantly reduce performance or cost-effectiveness. By analysing the MC-based DDS model through a semantic information perspective, we examine how correlations between the internalised particle concentration $(Y)$ and the particle concentration in the extracellular environment $(X)$ evolve under interventions. The final catalogue of results provides a quantitative basis for DDS design and optimisation, offering a method to determine optimal DDS parameter values under constraints such as chemical budget, desired effect and accuracy. Thus, the proposed framework can serve as a novel tool for guiding DDS design and optimisation.