Interpretable and Trustworthy AI for Early Arterial Fibrillation Diagnosis

Atrial fibrillation (AF) affects over 37 million people worldwide and can lead to stroke and heart failure. Early detection is difficult due to noisy signals and transient episodes. This PhD project "TrueBeat" aims to develop robust, interpretable, and energy-efficient AI models for early AF detection using ECG data, addressing clinical challenges through innovative AI and wearable tech. Develop signal processing techniques for ECG denoising and feature extraction Design and evaluate AI models including CNNs, LSTMs, Transformers, and classical Implement explainable AI (XAI) methods like SHAP and LIME for clinical interpretability Ensure model generalization across datasets and wearable devices Apply green AI strategies to improve computational efficiency and sustainability Develop prototype system and GUI for real-time AF detection Robust AI algorithms capable of early and accurate AF detection, explainable outputs facilitating clinical trust, and an energy-efficient system suitable for wearable devices, culminating in a prototype demonstrating real-time monitoring capabilities. Early and trustworthy AF diagnosis can drastically reduce stroke and heart failure risks. This interdisciplinary project will advance healthcare AI, supporting clinicians and patients with real-world applications and improved digital health technologies.

Bachelor’s and Post Graduate degree in Biomedical Engineering, Electrical/Electronic Engineering, Computer Science or related fields. Experience in Python/MATLAB, signal processing, or machine learning. Interest in AI for healthcare and wearable technologies. Strong analytical, programming, and communication skills. Demonstrated capability in research project delivery. Applicants not native in English must provide evidence of competency. Highly motivated with initiative and teamwork skills. Excellent academic and writing record. Desirable: Master’s degree in AI or biomedical engineering, experience with ML, healthcare data, and digital health. Prior scientific writing experience is an advantage.

Complete the online Application Form from the SETU website quoting the advert reference code SETU_2025_06_WSCH. Upload all supporting documents with your submission. Applications must be submitted via this route only. For application/admission queries, contact researchadmissions@setu.ie or call +353 (0)51 302883.