Seminars

Welcome to the BioHER Seminars page! Here, we foster a vibrant intellectual environment by hosting engaging discussions and presentations on the latest advancements in bioinformatics, health engineering, and related fields.

Our seminars bring together leading experts, researchers, and students from across our department, university, and the broader scientific community. These sessions are designed to share cutting-edge research, explore emerging technologies, and stimulate interdisciplinary collaboration. They offer a valuable platform for learning and professional growth. We regularly update this page with upcoming events, so check back often for details on our next sessions.

Abstract: The increasing complexity of solid tumors and colorectal liver metastases presents significant challenges in diagnostic imaging and treatment planning. Advanced computational imaging models based on intelligent systems could enhance the accuracy and interpretability of tumor characterization. By integrating explainable AI (XAI) techniques, the models provide both precise predictions and insights into the underlying decision-making processes, improving clinician trust and treatment outcomes. Digital twin technology allows creating dynamic, patient-specific simulations that can mirror tumor progression and response to therapies. These digital twins serve as personalized models, enabling more informed decisions in oncology by providing a robust platform for evaluating treatment strategies and predicting future tumor behavior. The aim is to improve the management of solid tumors, enhance personalized treatment planning, and ultimately, improve patient survival rates.

Abstract: Primary healthcare care data offers huge value in modelling disease and illness. However, this data holds extremely private information about individuals and privacy concerns continue to limit the wide-spread use of such data, both by public research institutions and by the private health-tech sector. One possible solution is the use of synthetic data which mimics the underlying correlational structure and distributions of real data but avoids many of the privacy concerns. Brunel University London has been working in a long-term collaboration with the Medicine and Health Regulatory Authority in the UK to construct a high-fidelity synthetic data generator using probabilistic models with complex underlying latent variable structures. This work has led to multiple releases of synthetic data on a number of diseases including covid and cardiovascular disease, which are available for state-of-the-art AI research. Two major issues that have arisen from our synthetic data work are issues with bias, even when working with comprehensive national data, and with concept drift where subsequent batches of data move away from current models and what impact this may have on regulation. In this talk I will discuss some of the key results of the collaboration: on our experiences of synthetic data generation, on the detection of bias and how to better represent the true underlying UK population, and how to handle concept drift when building models of healthcare data that evolves over time.

Abstract: Abel is more than a social robot: it is a multidisciplinary platform at the intersection of affective computing, artificial intelligence, robotics, psychology, and neuroscience. Designed as a hyper-realistic humanoid, Abel is capable of recognizing and expressing human emotions through multimodal sensing, e.g., facial expressions, speech prosody, body language, and physiological signals such as HRV, EDA, and respiration. Its cognitive architecture follows a sense–plan–act framework enriched by an internal emotional loop, allowing it to engage in believable and emotionally grounded interactions. By integrating advanced technologies such as wearable sensors, thermal imaging, remote photoplethysmography, and radar-based vital sign detection, Abel supports research into empathetic AI and embodied cognition. This platform aims to bridge the gap between machine functionality and emotional attunement, enabling adaptive, context-aware, and meaningful interactions. Abel is not only a research tool, but a visionary step toward human-centered AI systems that evolve, connect, and cooperate with us.

Abstract: The National Facility for Data Handling and Analysis at Human Technopole is tasked with providing analysis services in the areas of bioinformatics and image analysis to the Italian scientific community. To this aim, it needs analysis pipelines able to handle the wide variety of projects received, while ensuring automated and reproducible execution in a high-performance computing environment. This talk with describe how the National Facility is leveraging industry-standard tools and methodologies to implement a portfolio of powerful and flexible analysis pipelines that accommodate implement both standard projects as well as exploratory research. Thanks to the adoption of Nextflow and the nf-core standards, the pipelines are run seamlessly in different computational environments, and can easily be distributed in containerized form. This empowers users to become independent in performing further analysis projects. fulfilling the facility’s knowledge transfer mission. We will also describe an internally-developed object-oriented document generation toolkit that is used by the pipelines to generate complete and detailed analysis reports.

Abstract: Digital Health is revolutionizing healthcare with a greater focus on prevention rather than treatment. Its vision is to exploit technology advances to replace the existing point-of-care reactive system by a continuum-of-care, always-on proactive health system. This talk highlights the forces of change that will rapidly shape the future of health, including the increasing prevalence of digital health technologies, the use of data to personalize care and to benefit from the AI revolution we are witnessing today, and the changing role of patients in their care. Ultimately, the end game is to improve health outcomes, even eradicate certain diseases, while reducing the unit and total cost of care. Advancements in mobile computing, sensor technologies, and networking infrastructures are in the center of creating a paradigm shift in health management and delivery. These technologies enable innovative solutions for a wide range of needs, from monitoring chronic diseases and supporting healthy aging to providing emergency response and telemedicine services. By harnessing AI-driven analytics and therapeutics driven by real-time bio/social/behavioral sensor data fusion, we will be able to glean actionable insights—and even deliver interventions and actuations—from such large- and multi-scale, multimodal health data. We will examine past research and lessons learnt at the University of Florida including the role of wearables and smart homes in digital health. We'll explore how these technologies can enable graceful and safe aging- in-place, monitor patient health, provide personalized care, and support patients in managing their own health. The talk concludes with a discussion of the challenges and opportunities that lie ahead for digital health, and the role that mobility, sensing, and networking can play in shaping the future of health.

Abstract: The 21st century has witnessed major challenges caused by both COVID19 pandemic and aging society. In this talk, I will discuss the grand challenges and opportunities in AI for healthcare, and show some examples in areas such as AI Foundation Models, AI Implementation Science, and Metaverse. In AI Foundation Models, OpenAI LLMs primarily pretrain web-searched data. They are susceptible to hallucinated information and lack logical reasoning. Thus the healthcare LLMs require more domain expertise, patient specific data, logical reasoning to handle complex inferences, and computation and transparency for broad adoption in clinical settings. Working with Microsoft Accelerating Foundation Models Research, we developed the first retrieval augmented generation (RAG) solutions for clinical setting that augments LLMs with the most recent domain-specific medical knowledge. We then developed EHRAgent that augments external tools and medical knowledge to solve few-shot multi-tabular reasoning derived from EHRs (Electronic Health Records). EHRAgent formulates a clinical problem-solving process as an executable action sequence code plan with a code executor. Using interactive coding between the LLM agent and cod executor, the environment is feedback to improve code generation for tabular reasoning tasks in EHR. Compared to the state-of-the-art agent AutoGen, EHRAgent has 36% improvement, and has been published in Annual Meeting of the Association for Computational Linguistics (ACL2024) and the Empirical Methods in Natural Language Processing (EMNLP2024). Besides foundation model, our effort in AI Implementation Science has been accepted into AMIA 3-Tier AI Showcase, and our effort in real-time Digital Twin and Metaverse for rehabilitation have been accepted into Intelligent Reality conferences.