Thematic Session 25

THEMATIC SESSION #25

Micaela Morettini

Marche Polytechnic University (UnivPM), Italy

• m.morettini@univpm.it

Andrea Tura

CNR Institute of Neuroscience

• andrea.tura@cnr.it

Agnese Piersanti

Marche Polytechnic University (UnivPM), Italy

• a.piersanti@pm.univpm.it

Libera Lucia Del Giudice

Marche Polytechnic University (UnivPM), Italy

• l.l.delgiudice@pm.univpm.it

The rapid advancement of Generative Artificial Intelligence (AI) and Digital Twins (DT) hold promise for many biomedical engineering applications to foster health. Indeed, DTs, which are virtual replicas of biological systems, allow for real-time simulation, monitoring, and personalized treatment planning. However, such in-silico models require a huge amount and variety of data for their successful building. Moreover, real-world biomedical data often suffer from limitations such as data scarcity and variability, class imbalance and privacy concerns. In this context, generative AI techniques, of which Generative Adversarial Networks (GANs) are an example, may help improve DT technology, thus facilitating robust training for machine learning models, accelerating clinical research, and preserving patient privacy and limiting their discomfort, especially when data of interest rely on invasive procedures for their acquisition or comes from clinical trials.

This thematic session will explore the applications of generative AI and DT technology for synthetic data creation and signal generation in medicine and physiology. Specifically, it will address techniques and methodologies for generating high-fidelity synthetic biomedical images, biosignals, and patient profiles to address data scarcity and class imbalance. The session will also highlight the potential of synthetic datasets for model validation, simulation, and algorithm robustness testing. Participants are encouraged to present case studies, methodologies, and best practices that face challenges around the proposed thematic session.

Topics of interest include (but are not limited to):

• Synthetic biosignal and medical image generation;
• Data augmentation for model training;
• Applications of synthetic data in personalized medicine and digital twins;
• Privacy-preserving synthetic datasets;
• Applications based on Generative Adversarial Networks;
• Challenges and solutions in synthetic data validation and fidelity assessment;
• Clinical trials and real world data for Digital Twins.

Micaela Morettini holds a M.sc. in Biomedical Engineering (University of Bologna, 2008) and a PhD in “Electromagnetics and Bioengineering” (UnivPM, 2012). She is Associate Professor of Bioengineering at the Department of Information Engineering, UnivPM, where she is in charge of the DIABETES LAB. Her research interests include: in-silico modelling and digital twin technologies in physiology and medicine, wearable devices for health monitoring and disease management, biomedical signal processing, digital health and digital biomarkers, machine and deep learning applied to the development of clinical decision support systems. Main applications are in the diabetes, metabolism/immunometabolism, physical exercise, cardiovascular and respiratory fields. She currently is the coordinator of the Bachelor's and the Master's Degree in Biomedical Engineering at UnivPM. She is author of 77 journal papers and 80 conference proceedings.

Andrea Tura is Senior Research Scientist at the CNR Institute of Neuroscience. Currently, he is mainly focused on pancreatic beta-cell function and insulin secretion, insulin sensitivity, insulin clearance, glucose effectiveness and glucagon kinetics, especially in gestational diabetes and in type 2 diabetes. Other interests are flash/continuous glucose monitoring (also through non-invasive techniques) and assessment of glucose homeostasis in hemodialysis. Recently, he has also got interested in some complications of diabetes, especially sarcopenia. He is author of more than 200 articles in peer-reviewed scientific journals, of which more than 50 with primary role (first or last author). Current H-index (SCOPUS) is 43.

Agnese Piersanti got M.sc. in Biomedical Engineering in 2020, and PhD in Information Engineering at the Department of Information Engineering of UnivPM, in 2024, defending a thesis titled “Digital health technologies to improve diabetes prevention and optimize therapy: from model based approaches to feature based machine learning”. During the PhD, she spent an Erasmus+Traineeship period at the University of Southern California (Los Angeles, CA), for the development of an in-silico modeling approach to quantify insulin bioavailability. She had a one-year Post-Doc Fellowship on glucometabolic data analysis in gestational diabetes at the CNR Institute of Neuroscience (Padova, Italy). Currently, she is a Post-Doc Fellow at UnivPM, and her main research interests involve the development of bioengineering methodologies and low computational impact artificial intelligence algorithms for the study, management and treatment of major chronic diseases, such as diabetes.

Libera Lucia Del Giudice got M.sc. in Biomedical Engineering (cum laude) from UnivPM, in October 2023, with a thesis titled “Individual estimation of physiological parameters of the glucose-insulin regulatory system: a modeling approach for reduced-sampling oral glucose tolerance test data “. Since November 2023 she is a PhD Student at the Department of Information Engineering at UnivPM. Currently, her main research interests are in the development of mathematical models and methods, assisted by artificial intelligence strategies, in the field of metabolism and diabetes, with a particular focus on digital twins technology and generative artificial intelligence models for generating synthetic data in diabetes-related research.