Mechanics, Fracture, and Fragmentation of Thrombi: A Computational Study Review

Thrombosis drives myocardial infarction, acute ischemic stroke and pulmonary embolism, yet more than half of patients undergoing mechanical thrombectomy fail to achieve functional independence, and residual thrombus is detected in up to 91% of cases after catheter-directed thrombolysis. These persistent clinical failures reflect a fundamental gap: thrombus removal is not merely dissolution or extraction, but a multiphysics process governed by the fracture, deformation and fragmentation of a highly anisotropic composite material whose mechanical behavior remains poorly understood and insufficiently predicted. In this review, we examine physics-based and data-driven numerical models of thrombus mechanics, fracture, and fragmentation from a failure-oriented perspective and assess their potential clinical impact. A thrombus is a hierarchically organized composite in which a fibrin network scaffold, functionally differentiated platelet subpopulations and red blood cells interact across scales spanning six orders of magnitude, thus residual thrombus and fragment migration emerge as the fundamental challenges shared by all recanalization strategies from pharmacological thrombolysis to mechanical thrombectomy. We survey six major computational paradigms, including continuum-based approaches, lattice Boltzmann method, particle-based mesoscopic simulation, discrete fiber network models, multiscale coupling frameworks and artificial intelligence/machine learning-assisted methods, and trace the field's evolution over four decades from reaction kinetics to multiphysics fracture coupling. Finally, we delineate critical gaps in current simulation frameworks and argue that AI-driven surrogate models, physics-informed digital twins and generative virtual patient cohorts could transform thrombus simulation from a predominantly academic endeavor into a clinically actionable platform for real-time decision support. STATEMENT OF SIGNIFICANCE: Blood clots cause heart attacks, strokes, and pulmonary embolism, yet current treatments fail to fully remove clots in over half of cases. A key reason is that we still cannot predict how clots break apart during treatment. This review is the first to systematically examine thrombus simulation from a mechanical failure perspective, synthesizing six computational paradigms-from continuum mechanics to artificial intelligence-and tracing their four-decade evolution from modeling how clots form to predicting how they fail. By establishing residual thrombus and fragment migration as the shared unresolved challenges across all treatment strategies, we identify critical modeling gaps and propose AI-driven translational pathways, including digital twins and virtual patient cohorts, that could transform clot simulation into a real-time clinical decision-support tool.