Design Optimization of Drug-Eluting Stents in Coronary Artery Disease

Abstract

Drug-eluting stents (DES) have revolutionised percutaneous coronary intervention by reducing in-stent restenosis rates from 20-30% with bare-metal stents to 5-10% through local antiproliferative drug delivery. However, DES design remains a multi-objective optimisation problem: the stent must provide sufficient radial strength to scaffold the artery, flexibility to navigate tortuous vessels, uniform drug distribution to prevent neointimal hyperplasia, and biocompatibility to minimise late stent thrombosis. Current DES designs rely on iterative empirical testing -- a costly process requiring 3-5 years and USD 50-100 million per design cycle. Computational design optimisation using finite element analysis (FEA), computational fluid dynamics (CFD), drug transport modelling, and machine learning-guided multi-objective optimisation can accelerate DES development by virtually screening thousands of design variants before physical prototyping. We present the Drug-Eluting Stent Optimisation Framework (DESOF), evaluating five computational design approaches - FEA-based structural optimisation, CFD haemodynamic analysis, pharmacokinetic drug release modelling, topology optimisation, and ML-guided multi-objective optimisation -- across four DES design parameters (strut geometry, polymer coating architecture, drug formulation, and deployment mechanics). Our DES Design Optimisation Score (DDOS) measures structural performance, drug delivery uniformity, haemodynamic compatibility, fatigue life prediction, and manufacturing feasibility. ML-guided multi-objective optimisation achieves the highest DDOS (0.926) by simultaneously optimising all design parameters within manufacturing constraints, while topology optimisation achieves the highest structural efficiency (0.960) through algorithmically generated strut geometries that minimise material while maximising radial strength.