XAI on Biosignals for CDSS
ACM KDD, Sunday 25 August 2024 10:00 AM - 01:00 PM, Barcelona Spain
Tell Me More

Abstract

Explainable AI on Biosignals for Clinical Decision Support.

  • Background

    Deep learning has proven effective in several areas, including computer vision, natural language processing, and disease prediction, which can support clinicians in making decisions along the clinical pathway. However, in order to successfully integrate these algorithms into clinical practice, it is important that their decision-making processes are transparent, explainable, and interpretable.

  • Targeted XAI

    Firstly, this tutorial will introduce targeted eXplainable Artificial Intelligence (XAI) methods to address the urgent need for explainability of deep learning in healthcare applications. In particular, it focuses on algorithms for raw biosignals without prior feature extraction that enable medical diagnoses, specifically electrocardiograms (ECG) – stemming from the heart – and electroencephalograms (EEG) representing the electrical activity of the brain.

  • Workflow

    Secondly, participants are provided with a comprehensive workflow that includes both data processing and an introduction to relevant network architectures. Subsequently, various XAI methods are described and it is shown, how the resulting relevance attributions can be visualized on biosignals.

  • Real World Use Case

    Finally, two compelling real-world use cases are presented that demonstrate the effectiveness of XAI in analyzing ECG and EEG signals for disease prediction and sleep classification, respectively. In summary, the tutorial will provide the skills required for gaining insight into the decision process of deep neural networks processing authentic clinical biosignal data.

  • Be Part
    Of Our
    Workshop!

Slides

Presentation slides for our workshop.

Workhsop Slides

Full slide deck

Jupyter Notebooks

Jupyter Notebooks

Google drive containing data and jupyter notebooks for the workshop.

Our Amazing Team

Jun.-Prof. Dr. Anne-Christin Hauschild

Anne-Christin Hauschild is leading the Clinical Decision Support Systems Group and deputy head of the Department of Medical Informatics at the University Medical Center Göttingen. Her group develops complex AI algorithms, such as tailored Explainable AI methods for biomedical applications increase the causal understanding of decision processes of predictive models. She finished her PhD with Prof. Dr. Jan Baumbach at the Max-Planck Institute for Informatics in Saarbrücken in 2016. Thereafter, she worked as a Research Fellow at the Princes Margaret Cancer Center and subsequently at the Center for Addiction and Mental Health in Toronto (Ca). In 2019 joined the Bioinformatics group at the Philipps University of Marburg as a junior group leader for Medical Data Science.

Dr. Nicolai Spicher

Nicolai Spicher is leading the Biosignal Processing Group at University Medical Center Göttingen. His research interests include signal processing and machine learning for cardiovascular and multimodal applications. He is Adcom member of the IEEE Engineering in Medicine and Biology Society (EMBS) and serves as chair of German/Austrian/Swiss EMBS chapter. Dr. Spicher received his Ph.D. degree from the Erwin L. Hahn Institute for MRI and was a postdoc at the the Peter L. Reichertz Institute for Medical Informatics at TU Brauschweig and Hannover Medical School.

Dr. Theresa Bender

Theresa Bender is a Postdoc at the Department of Medical Informatics at the University Medical Center Göttingen, Germany. She recently received a doctoral degree from the GAUSS Programme in Computer Science, analyzing the trustworthiness of deep learning models for 12-lead ECGs. She received a B.Sc. and consecutive M.Sc. in Applied Informatics with focus on Healthcare Informatics at Georg-August University Göttingen, Germany. As a member of the Biosignal Processing Group she is now focusing on post-hoc explainability for ECG and EEG data.

Jacqueline Michelle Metsch

Jacqueline Michelle Metsch is a PhD student at the Department of Medical Informatics at the University Medical Center in Göttingen. She earned her B.Sc. in Mathematics from the Justus-Liebig-University Gießen and her M.Sc. in Data Science from the Philips-University Marburg. As a member of the Clinical Decision Support Systems (CDSS) Group she is working on building trust in AI as CDSS through the use of XAI.

Philip Hempel

Philip Hempel is a PhD student at the Department of Medical Informatics at the University Medical Center in Göttingen. He got his B.Sc and M.Sc in Biology at the Technical-University Braunschweig and has a state examination as paramedic accompanied by 11 years of working experience. In the biosignal processing group he links the performance of AI-ECG models to medical textbook knowledge to facilitate their use in clinical practice.

Miriam Cindy Maurer

Miriam Cindy Maurer is a PhD student at the Department of Medical Informatics at the University Medical Center in Göttingen. She earned her B.Sc. in Mathematics and Economics from the Ruprecht-Karls University Heidelberg and her M.Sc. in Applied Statistics from the Georg-August University Göttingen. As a member of the Clinical Decision Support Systems (CDSS) Group, she focuses on Graph Neural Networks applied to ECG data, with a special emphasis on XAI.

Philip Zaschke

Philip Zaschke studied Medical Computer Science at Bachelor's and Master's level at the Department of Medical Informatics, University Medical Center Göttingen, Germany. In the research groups collaborative clinical research and biosignal processing he could gain particular expertise in deep learning, high performance computing and FAIR infrastructure development. For EEG sleep data, he focuses on the topics high performance infrastructure development and post-hoc explainability of neural networks.