Prediction of peripheral arterial disease through non-invasive diagnostic approach

Abstract

Peripheral arterial disease (PAD) is a cardiovascular condition caused by arterial blockages and poor blood circulation, increasing the risk of severe complications such as stroke, heart attack, and limb ischemia. Early and accurate detection is essential to prevent disease progression and improve patient outcomes. This study introduces a non-invasive diagnostic method using laser doppler flowmetry (LDF), electrocardiography (ECG), and photoplethysmography (PPG) to assess vascular health. LDF measures microvascular blood flow, ECG evaluates heart rate variability, and PPG captures pulse waveform characteristics. Key physiological features such as blood flow variability, pulse transit time, and hemodynamic responses are extracted and analyzed using machine learning. Random forest and XGBoost models are employed and combined using ensemble learning to classify individuals into non-PAD, moderate PAD, and severe PAD categories. A comparative evaluation shows that the ensemble model delivers superior classification accuracy. This integrated system offers a fast, reliable screening tool that supports early PAD detection and intervention. By combining multimodal signal analysis with machine learning, the approach enhances diagnostic precision and provides a scalable solution for preventive cardiovascular care.