Probabilistic clinical studies and deterministic physiology: building a Monte Carlo bridge with the "Frank" project

To resolve the apparent paradox between deterministic physiological laws and probabilistic clinical outcomes by framing clinical variability as epistemic uncertainty within a complex, non-linear system, demonstrating how deterministic mechanisms can account for empirical distributions and outcome heterogeneity.

We developed the "Frank" project, a 0-dimensional computational agent-based model (ABM) of human cardiovascular hemodynamics implemented in Julia (Agents.jl). Using Monte Carlo simulations (n=100), myocardial infarction was modeled by varying initial conditions, specifically sampling the infarction area from a Gaussian distribution along with systemic compliance. Cardiogenic shock was defined by a systolic arterial pressure <90mmHg and a cardiac index <2.2L/min/m2, with parameters evaluated via grid search against literature incidence benchmarks (5-10%).

The simulation produced hemodynamically stable, physiologically sound boundary parameters and pressure-volume loops. When initialized with a mean infarction area of 35±5% and systemic compliance of 1.5±0.3mL/mmHg, the model yielded a 7\% incidence of cardiogenic shock, closely matching empirical data. A strong linear inverse correlation was observed between infarction area and post-infarction systolic blood pressure (mean 93.96mmHg).

Stochastic-appearing clinical trial outcomes can emerge from deterministic physiological systems characterized by sensitivity to initial parameters and hidden variables. The "Frank" framework establishes a Monte Carlo bridge between mechanistic physiology and epidemiological statistics, offering an approach to explore outcome heterogeneity and refine clinical trial interpretation and patient stratification.