From Biological Mechanisms to Causal Inference: Quantitative Systems Pharmacology and Causal Frameworks in Drug...

Digital patients and in silico clinical trials offer new opportunities to advance pharmacological research, clinical development, and regulatory decision-making by enabling the simulation of treatment effects prior to testing in patients. Among the most promising of these opportunities is the combined use of quantitative systems pharmacology (QSP) and causal inference (CI), two complementary paradigms that, together, address different but interlocking aspects of causal reasoning in drug development: one grounded in mechanistic simulation, the other in data-driven estimation. Mechanistic quantitative systems pharmacology (QSP) models generate physiologically interpretable predictions by encoding biological knowledge-such as compartmental flows, receptor kinetics, and feedback loops-typically in systems of differential equations. As such, QSP addresses the question: given specified mechanistic assumptions, how does the system behave under intervention? Complementing this mechanistic lens, causal inference approaches start from observed data and estimate the effects of interventions using formal causal frameworks and comparative outcome analysis. These methods explicitly address confounding, multicausality, and clinical heterogeneity, asking instead: given the data, what are the causal effects of intervening in the real world? These paradigms reflect fundamentally different representations of causality-mechanism-driven versus data-driven-two complementary lenses on the same underlying goal of understanding and predicting intervention effects. This distinction raises a central question: how can causality be consistently represented, inferred, and validated across modeling approaches, and what is required to integrate them to support robust and credible decision-making?