15.5.2024. 18:30h, Faculty of Mathematics (online)

Machine learning in medicine: Sepsis prediction and antibiotic resistance prediction

Prof. Dr. Karsten Borgwardt

Max Planck Institute of Biochemistry, Department of Machine Learning and Systems Biology, Germany

    Video: Recorded lecture (MP4, 65min, 95MB)

A meeting of the Bioinformatics seminar will be held on Wednesday, May 15th, starting at 18:30, in online classroom. Teachers and students of doctoral and master studies in computer science, mathematics, biology and other related disciplines are invited to join us.

Abstract

Sepsis is a major cause of mortality in intensive care units around the world. If recognized early, it can often be treated successfully, but early prediction of sepsis is an extremely difficult task in clinical practice. The data wealth from intensive care units that is increasingly becoming available for research now allows to study this problem of predicting sepsis using machine learning and data mining approaches. In this talk, I will describe our efforts towards data-driven early recognition of sepsis and the related problem of antibiotic resistance prediction.

Lecturer

Karsten Borgwardt is Director of the Department of Machine Learning and Systems Biology at the Max Planck Institute of Biochemistry in Martinsried, Germany since February 2023. His work won several awards, including the 1 million Euro Krupp Award for Young Professors in 2013 and a Starting Grant 2014 from the ERC-backup scheme of the Swiss National Science Foundation. Prof. Borgwardt has been leading large national and international research consortia, including the “Personalized Swiss Sepsis Study” (2018-2023) and the subsequent National Data Stream on infection-related outcomes in Swiss ICUs (2022-2023), and two Marie Curie Innovative Training Networks on Machine Learning in Medicine (2013-2016 and 2019-2022).

Seminar

The organizer of the seminar is BIRBI. The heads of the seminar are Prof. dr Nataša Pržulj and dr Jovana Kovačević.

24.4.2024. 18:15h, Faculty of Mathematics (online)

From Climate to Healthcare: Adapting Polar Diagrams for Biomedical and Machine Learning Applications

Aleksandar Anžel

Centre for Artificial Intelligence In Public Health Research, Robert Koch Institute, Berlin, Germany

    Video: Recorded lecture (MP4, 52min, 62MB)

A meeting of the Bioinformatics seminar will be held on Wednesday, April 24th, starting at 18:15, in online classroom. Teachers and students of doctoral and master studies in computer science, mathematics, biology and other related disciplines are invited to join us.

Abstract

Evaluating the performance of multiple complex models, such as those found in biology, medicine, climatology, and machine learning, using conventional approaches is often challenging when using various evaluation metrics simultaneously. The traditional approach, which relies on presenting multi-model evaluation scores in the table, presents an obstacle when determining the similarities between the models and the order of performance. By combining statistics, information theory, and data visualization, juxtaposed Taylor and Mutual Information Diagrams permit users to track and summarize the performance of one model or a collection of different models. To uncover linear and nonlinear relationships between models, users may visualize one or both charts.

This presentation will delve into the mathematical foundation of both the Taylor and Mutual Information Diagrams, highlighting their distinctions and similarities. Attendees will have the opportunity to view the resulting diagrams generated by the new library named polar-diagrams. This library offers the first publicly available implementation of the Mutual Information Diagram and the first interactive implementation of the Taylor Diagram. Furthermore, the presentation will discuss the additional features recently integrated into both diagrams, enabling the visualization of temporality or specific scalar model attributes like uncertainty. These concepts will be illustrated using sample datasets from diverse fields, including climatology, biomedicine, public health, and machine learning.

Lecturer

Dr. Aleksandar Anžel is a Postdoctoral Researcher at the Robert Koch Institute, where he actively contributes to pioneering research endeavors within the public health domain. His specialization lies in optimizing visualization techniques for AI models, developing early-detection surveillance systems, and improving existing computational workflows, among others.

Aleksandar completed his Ph.D. at Philipps-Universität Marburg, where he focused his research on improving existing and developing new bioinformatics pipelines and tools that leverage machine learning and data science methodologies. His work also revolved around developing, evaluating, and visualizing automated workflows for information storage systems utilizing molecular storage media like DNA. Moreover, he worked on improving existing and developing novel techniques for analyzing and visualizing high-dimensional multi-modal data sets, including temporal multi-omics data. During his Ph.D. Aleksandar was also employed as Technical Lead at eMedicals Healthtech GmbH, where he oversaw the development of the kidi project platform. The project, designed as Software as a Medical Device (Digital Health Application (DiGA)), was developed following rigorous industry standards to ensure patient safety and regulatory compliance.

Furthermore, Aleksandar holds a Master's degree in Mathematics with a specialization in Computer Science and Informatics from the Faculty of Mathematics at the University of Belgrade. He graduated with a Master's Thesis focusing on "Determining Protein N-glycosylation with Machine Learning Methods".

More information about Aleksandar can be found at https://aanzel.github.io/

Seminar

The organizer of the seminar is BIRBI. The heads of the seminar are Prof. dr Nataša Pržulj and dr Jovana Kovačević.

10.4.2024. 18:15h, Faculty of Mathematics (online)

Using random walks to explore complex networks

Alexandre V. Morozov

Department of Physics and Astronomy and Center for Quantitative Biology, Rutgers University, USA

    Video: Recorded lecture (MP4, 54min, 68MB)

A meeting of the Bioinformatics seminar will be held on Wednesday, April 10th, starting at 18:15, in online classroom. Teachers and students of doctoral and master studies in computer science, mathematics, biology and other related disciplines are invited to join us.

Abstract

Large-scale networks represent a broad spectrum of systems in nature, science, technology, and human and animal societies. The complexity of these networks makes predictions of their properties a challenging task. I will describe a novel computational methodology, based on random walks, for the inference of both local and global properties of complex networks. I will show that our formalism yields reliable estimates of key network properties, such as its size, after only a small fraction of network nodes has been explored. Furthermore, I will introduce a novel algorithm for partitioning network nodes into non-overlapping communities - a key step in revealing network modularity and hierarchical organization. Thus, non-ergodic random walk trajectories help reveal modular organization and global structure of complex networks.

[1] Kion-Crosby, W.B. & Morozov, A.V. (2018). Rapid Bayesian inference of global network statistics using random walks. Phys. Rev. Lett. 121, 038301.

[2] Ballal, A., Kion-Crosby, W.B. & Morozov, A.V. (2022). Network community detection and clustering with random walks. Phys. Rev. Res. 4, 043117.

[3] Yu, J. & Morozov, A.V. (2024). An adaptive Bayesian approach to gradient-free global optimization. New J. Phys. 26, 023027.

Lecturer

Alexandre V. Morozov received his Ph.D. in Physics from the University of Washington, Seattle in 2003. From 2003 to 2007 he was a post-doctoral fellow at the Center for Studies in Physics and Biology, Rockefeller University, New York. In 2007 he joined the Department of Physics and Astronomy at Rutgers University, where he is now Professor and Director of the Center for Quantitative Biology. In 2009, he was a recipient of an Alfred P. Sloan Research Fellowship. His current research interests include non-equilibrium statistical mechanics, machine learning, biological physics, and evolutionary modeling.

Seminar

The organizer of the seminar is BIRBI. The heads of the seminar are Prof. dr Nataša Pržulj and dr Jovana Kovačević.