A data-driven framework to identify restenosis-prone regions in femoral arteries from geometric and inflow waveform parameters.

Haemodynamic indices derived from computational fluid dynamics (CFD), such as time-averaged wall shear stress (TAWSS) and oscillatory shear index (OSI), are closely associated with restenosis risk in peripheral arterial disease (PAD). We present a machine learning-reduced order modelling (ML-ROM) framework to predict critical, restenosis-prone, haemodynamic regions accounting for both vessel geometries and inlet flow waveforms. We generated 500 parameterized virtual femoral-artery geometries, each paired with a synthetic physiologically realistic inflow waveform. CFD was used to obtain the wall shear stress (WSS) distribution, from which TAWSS and OSI were computed. Critical regions were then defined by applying different thresholding criteria based on TAWSS and OSI, as found in the literature. These were then used to construct ROMs using Proper Orthogonal Decomposition (POD), and to train ML models from which the critical regions could be reconstructed. Four ML architectures were explored: a geometric-parameter-only model and three alternatives that also incorporate inflow waveform information via Fourier coefficients, long short-term memory (LSTM), and convolutional neural networks (CNN). The Fourier models achieved the highest predictive performance, with the median values of balanced accuracy (BA) exceeding 0.92 across all critical-region definitions. The model also offered a substantial speed-up ratio of the order of 109 compared with traditional CFD.