SimEval: a toolbox of software and interactive notebooks for credibility assessment of medical device modeling and simulation

SimEval is an open-source toolbox of Python routines and interactive Jupyter notebooks supporting credibility assessment of medical device computational models. Release 1.0, focused on verification, includes: (1) a Jupyter notebook covering end-to-end code verification; (2) a Python module for computing key calculation verification metrics with automated decision logic; and (3) a study investigating code verification of physics-informed neural networks (PINNs), including a proposed PINN-specific code verification workflow. SimEval helps bridge the gap between advanced simulation software and model credibility methods; this release supports increased verification efficiency and rigor which is especially important as generative AI is increasingly enabling rapid development of simulation software that may contain subtle errors. SimEval is available at https://github.com/OSEL-DAM/SimEval.