NLP and AI For Health

Title: Artificial Intelligence Applied to Psychiatry: Uses and Limitations of AI in Diagnosing Mental Disorders

Abstract 

Mental disorders present a significant challenge in healthcare due to the lack of specific biomedical tests and the need for close, continuous monitoring to ensure accurate diagnosis. Recent advances in AI, particularly in Natural Language Processing, show great potential in assessing these conditions. However, there are still significant challenges to translating this potential into the reality of psychiatric practice. In this presentation, I will discuss the work conducted at UFRN on discourse analysis in clinical settings for various disorders. Topics will include: the state of the art in detecting cognitive neuropathies, the search for descriptors that enable the development of explainable diagnostic systems, the construction of neuroscientific models that link speech patterns to neural mechanisms, and the challenges in creating technology that integrates seamlessly with clinical practice.

About the Speaker

Dr. César Rennó-Costa is an Assistant Professor of Bioinformatics and the Director of the Bioinformatics Multidisciplinary Environment at the Digital Metropolis Institute, Federal University of Rio Grande do Norte (UFRN), Natal, Brazil. His research background lies in computational neuroscience, particularly in modeling biological neural networks, focusing on hippocampal circuits. He has participated in several collaborative projects with US and European institutions. More recently, his work has expanded into applied medical research, including developing digital systems utilizing AI and data science to support public health agencies, mainly focusing on behavioral sciences and computational psychiatry through natural language processing techniques. He currently leads an international network financed by the Newton Fund aimed at developing computational methods for characterizing and identifying mental disorders.

Title: Collaboration Opportunities in Digital Health at the Digital Metropolis Institute in Brazil

Abstract 

This presentation highlights the collaborative opportunities in digital health at the Digital Metropolis Institute (IMD) in Brazil, an innovation hub at the Federal University of Rio Grande do Norte (UFRN). I will introduce the institute’s structure, including its bioinformatics center, and discuss key research and educational projects. Additionally, I will present our work with the State Health Authority on developing an automated epidemiology system, as well as our innovation initiatives with local healthcare providers and stakeholders.

About the Speaker

Dr. Renan C. Moioli is an assistant professor at the Digital Metropolis Institute (IMD) from the Federal University of Rio Grande do Norte, Brazil, where he is also part of the Graduate Program in Bioinformatics. He has a D. Phil. in Cognitive Science from the University of Sussex (2013), UK, working at the Centre for Computational Neuroscience and Robotics (CCNR). From 2013 to 2018 he was a research fellow at the Edmond and Lily Safra International Institute of Neuroscience, Brazil, working on brain-machine interfaces. From 2019 to 2022, he was a UK Royal Society Newton Advanced Fellow, working in close collaboration with Heriot-Watt's Robotics Laboratory. His research interests are in the fields of computational intelligence and bioinformatics.

Title: "Sifílis Não" Project: the largest initiative to combat syphilis in the context of global health.

Abstract 

The “Syphilis No Project” is one of the largest global initiatives to combat syphilis, the result of a collaboration between the Laboratory of Technological Innovation in Health (LAIS) at the Federal University of Rio Grande do Norte (UFRN), the Brazilian Ministry of Health, and the Pan American Health Organization (PAHO/WHO). The central goal of the project is to reduce cases of acquired syphilis and syphilis in pregnant women in Brazil through four key areas of focus: management and governance, epidemiological surveillance, comprehensive care, and strengthening education and communication. In this lecture, I will discuss the project’s actions toward eliminating congenital syphilis, with an emphasis on prevention and early diagnosis in pregnant women, as well as the integration of surveillance and healthcare efforts within the Brazilian Unified Health System (SUS) and the development of academic and medical research that seeks new approaches to controlling this sexually transmitted infection in Brazil.

About the Speaker

Dr. Ana Isabela is a biomedical scientist with a PhD in Cellular and Molecular Biology from the Ribeirão Preto Medical School at the University of São Paulo (USP). She is currently a substitute lecturer at the Health Sciences Centre at the Federal University of Rio Grande do Norte (UFRN) and a researcher at the Health Innovation Laboratory (LAIS) at UFRN. During the COVID-19 pandemic, she served as a technical consultant, strengthening epidemiological surveillance and laboratory diagnosis initiatives for the State Secretariat of Public Health of Rio Grande do Norte (SESAP-RN), in a project coordinated by the Ministry of Health (MS) and funded by the Pan American Health Organization (PAHO) and the World Health Organization (WHO). Professor Ana Isabela frequently acts as a consultant for the National Validation Teams of the Ministry of Health, conducting technical visits to Brazilian municipalities to assess and evaluate the implementation of clinical protocols and therapeutic guidelines. As a researcher at LAIS, she contributes to the development of new diagnostic methods and to the integration between surveillance and healthcare within the scope of the "Syphilis No" project, supported by the Ministry of Health and PAHO, which is the largest syphilis prevention project in the context of global health. She has extensive experience in biomedical methodologies and health systems.

Title: Creating data assets for collaborative partnerships: the opportunities in the healthcare domain

Abstract 

The presentation is about an overview of problems challenges in the healthcare domain and how different types of linked data has the potential to answer different questions, but ultimately benefiting individuals and populations. Whilst the core story is around structured coded data, there will also be a posing of the question on how we bring unstructured data together in the application context.

About the Speaker

Professor Umesh Kadam is a clinician and an epidemiologist with a long history of working with large datasets from UK, Sweden and the Netherlands. His main interest is in the  use of routinely collected clinical and healthcare data to improve population and patient care. His research has been funded by HDRUK, MRC, Wellcome Trust and NIHR, and covers both innovation in methods as well as applications to the real-world environment.

Title: Towards Ethical Collaboration Between Human and AI 

Abstract 

The talk will introduce several data-driven approaches to address bias and privacy concerns in Healthcare AI, such as filtering training data based on information density and ensuring privacy through data fragmentation. It will also delve into human-centered strategies, focusing on how large language models (LLMs) can facilitate healthcare data mining by experts and the integration of uncertainty-aware AI in clinical decisions. We will also present our investigation of human privacy-preserving behaviors in chatbot interactions. Finally, the talk will explore how reinforcement learning can help language models better leverage human expertise.

About the Speaker

Dr. Julia Ive is an Associate Professor in Artificial Intelligence applied to Healthcare at UCL. Her research focuses on data-centric and human-centered approaches to large language models (LLMs) to drive responsible healthcare innovation. Previously, at Queen Mary and Imperial, she worked on AI algorithms for phenotyping, automated report generation, synthetic text generation to address data sparsity in mental health, as well as uncertainty-aware algorithms for clinical decision-making. 

Title: AI in Basic and Clinical Neuroscience 

Abstract 

Traditionally, neural networks have been used for data analysis and as models of the mind and the brain. These two areas have both made historically significant contributions. For example, connectionism as a model of the brain has helped cognitive psychologists to understand many computational principles in language acquisition, memory and control of action. Although modern AI inherited many fundamental structures and features in conventional neural network models, its current applications in neuroscience have primarily been data analysis, e.g. for MRI data. In this talk, I will show how modern AI can contribute to both data analysis in neuroscience and also help theorising computational principals implemented by the brain. Examples shown in this presentation will cover both basic and clinical neuroscience including computational linguistic and speech recognition.  

About the Speaker

Professor Li Su is Professor of Neuroimaging and the Head of the Artificial Intelligence and Computational Neuroscience Group. He has a joint appointment at both the Department of Psychiatry, University of Cambridge and the Neuroscience Institute, University of Sheffield. He is a Fellow and Tutor at Clare Hall, University of Cambridge, AI for Health panel member at UK Research and Innovation (UKRI), a Specialist Committee Member at National Institute for Health and Care Excellence (NICE) and member of the Steering group at National BRC Imaging Network. He is the regional lead for the Dementias Platform UK (DPUK) imaging network and the regional co-lead for the National Network for the Application of Data Science and AI to Dementia Research (DEMON), member of Alzheimer's Society's Research Strategy Council and member of Grant Review Committee for the Lewy Body Society. He was leading the ARUK East Network Centre until 2022. He has received numerous awards including Alzheimer’s Research UK Senior Research Fellowship Award (2017), International Psychogeriatric Association Junior Research Award in Psychogeriatrics (2016), and International College of Geriatric Psychoneuropharmacology Junior Investigator Award (2015). 

Title: TBC 

Abstract 

About the Speaker

Professor Edward Watkins is Professor of Experimental and Applied Clinical Psychology at the School of Psychology, University of Exeter. He is a chartered clinical psychologist specialising in cognitive-behavioural therapy and director of the Mood Disorders Centre at the University of Exeter. His work focuses on the psychological understanding, treatment and prevention of worry, rumination, anxiety, and depression using face-to-face and digital approaches, across the lifespan from adolescence into young adults and beyond. His research has been funded by the Wellcome Trust, UK Medical Research Council, European Commission and US NIMH. He currently leads Nurture-U a large UKRI-funded project to develop and evaluate a stepped change whole university approach to improve student mental health, and a Wellcome Trust Mental Health award focusing on understanding how cognitive-behavioural therapy changes repetitive negative thought. 

Title: Multilingual LLM Evaluation in Practical Settings

Abstract 

Large language models (LLMs) are increasingly used in a variety of applications across the globe but do not provide equal utility across languages. In this talk, I will discuss multilingual evaluation of LLMs in two practical settings: conversational instruction-following and usage of quantized models. For the first part, I will focus on a specific aspect of multilingual conversational ability where errors result in a jarring user experience: generating text in the user’s desired language. I will describe a new benchmark and evaluation of a range of LLMs. We find that even the strongest models exhibit language confusion, i.e., they fail to consistently respond in the correct language. I will discuss what affects language confusion, how to mitigate it, and ongoing extensions. In the second part, I will discuss the first evaluation study of quantized multilingual LLMs across languages. We find that automatic metrics severely underestimate the negative impact of quantization and that human evaluation—which has been neglected by prior studies—is key to revealing harmful effects. Overall, I highlight limitations of multilingual LLMs and challenges of real-world multilingual evaluation.

About the Speaker

Sebastian Ruder is a research scientist based in Berlin, Germany. He leads the Multilinguality team at Cohere whose mission is to improve the multilingual capabilities of Cohere's large language models (LLMs). Before that he was a research scientist at Google DeepMind. He completed his PhD in Natural Language Processing (NLP) at the Insight Research Centre for Data Analytics, while working as a research scientist at Dublin-based text analytics startup AYLIEN. Previously, he studied Computational Linguistics at the University of Heidelberg, Germany and at Trinity College, Dublin.