Treffer: Translation of a cardiorespiratory model of patients under mechanical ventilation

Mechanical ventilation management in intensive care units relies heavily on accurate, realtime physiological models, but existing Simulink‐based implementations impose high computational costs, limited traceability, and proprietary dependencies. This work presents the translation and standalone reimplementation of two key subsystems, Respiratory and Cardiovascular, of a pre-existing cardiorespiratory model into pure MATLAB code. A statespace formulation was adopted to decouple the model dynamics from Simulink solvers and facilitate future translation to open-source environments. Custom integration loops employing explicit Euler, Heun, and fourth-order Runge–Kutta (RK4) methods were developed and evaluated across a range of time steps (dt). The different simulation configurations were quantified using Percentage Prediction Error (% PE) against a high-fidelity RK4 reference (dt = 0.001). Computational performance was measured via repeated tic-toc evaluations, and an optimization index (Precision/Time) was introduced to rank dt-method configurations. Results confirmed the theoretical sensitivity of accuracy to dt,but differences among integration methods were minor and, in some cases, unexpected. When compared with the original Simulink output, the Cardiovascular System achieved strong waveform agreement (%PE ≈ 10 %, Pearson ≈ 0.75), while the Respiratory System exhibited larger offsets (%PE ≈ 23 %) due to omitted controller loops. These findings demonstrate that a state-space-based MATLAB reimplementation can deliver near-equivalent physiological fidelity with improved code clarity and licensing independence. Future work would complete the translation of missing control loops and refine vectorized implementations to reduce computation time further, easing the way for an integrated, open-source cardiorespiratory modeling platform.