ResearchGroup

Description: The research candidate will join the Biomedical Signal Interpretation and Computational Simulation (BSICoS) group (https://bsicos.i3a.es/), belonging to the University of Zaragoza (UNIZAR), CIBER on Biomedical Engineering, Biomaterials and Nanomedicine (CIBER-BBN) and Aragon Institute for Health Research (IIS Aragon), where Dr. Juan Pablo Martínez is one of the faculty members. The research group is a multidisciplinary team formed by 34 researchers: 12 faculty members (including 6 faculty at UNIZAR, 1 ARAID researcher and 2 researchers from Centro Universitario de la Defensa), 8 post- doctoral researchers and 16 PhD students. In addition, the group has an extensive network of international collaborations with groups at a wide range of Universities, some of whose researchers have been actively contributing to the main activities of the group in the past (most of them as former PhD students) and maintain their collaboration.

ResearchLine

Description: Population-based Risk Stratification and Prediction of Life-Threatening Ventricular Arrhythmias through Transfer Learning Sudden cardiac death (SCD) occurs without warning in 50% of cases, often due to life-threatening ventricular arrhythmias (LTVA). Early, non-invasive electrogram (ECG) analysis may offer cost-effective SCD stratification. Past research extracted ECG risk indices based on electrophysiology, supported by experiments. Neural networks (NN) may also perform ECG analysis, identifying vital relevant features for risk stratification. Labelled ECG data remains a limitation, but pretraining NN on unlabelled data enhances performance. This project aims to enhance NN performance in LTVA classification by building upon pretrained models optimized for simpler tasks using unlabelled ECG signals. Our primary goal is to develop such a pretrained model, using transfer learning to fine- tune high-performing models for LTVA risk prediction in general and specific populations.

Others

Description: To align with the MSCA Green Charter, the project will prioritize environmentally sustainable practices. This includes using energy-efficient computing infrastructure for training and deploying the neural network, minimizing the environmental impact of data collection and processing, and adopting a green and sustainable approach to research, such as using renewable energy sources for computing tasks whenever possible. We will adhere to ethical guidelines and data privacy regulations: ensuring patient data confidentiality, informed consent, secure data handling and GDPR compliance. As the project involves artificial intelligence and health data, we will also adhere to national and regional policies on the use of AI in healthcare. Special attention will be paid to identify and reduce any possible bias in the training data. Open science practices will be encouraged, by sharing data and code when possible. Collaboration with other researchers and institutions and interdisciplinary cooperation is a common practice in our research group, and so will be in this project. By addressing these aspects, the research project will also contribute to a more sustainable, ethical, and responsible application of technology in the medical domain.

CareerDevelopment

Objectives

• Future postdoc employment at the University
• International mobility

Expected Outputs

• PhD defense
• Publication of articles
• Attendance to international conferences

Training Skills

• Grant Writing
• Communication to the civil society and public engagement
• Open Science
• Ethics and Research Integrity
• Teaching
• Intellectual Property
• Leadership
• Responsible use of Artificial Intelligence
• Spanish