-
Luna Bitar (Universität Hamburg)
Autoimmune Hepatitis (AIH) and Drug-Induced Liver Injury (DILI) cannot be reliably distinguished by pathologists from liver biopsy images alone. We investigate whether deep learning can identify image patterns that differentiate AIH from DILI in digitized Whole Slide Images (WSIs).
Go to contribution page
Patch-level features were extracted using the UNI foundation model and aggregated into slide-level predictions... -
Dr Victor Dudarev (Ruhr University Bochum)Poster and Lightning Talk
High-throughput materials research generates large volumes of heterogeneous experimental data that must be integrated to support reproducible analysis and data-driven discovery. We present a Research Software Engineering approach implemented in the MatInf Research Data Management System (RDMS) for integrating multimodal screening data obtained from thin-film materials libraries. The...
Go to contribution page -
Anja Witte (Institute of Medical Systems Bioinformatics, Center for Biomedical AI (bAIome), Center for Molecular Neurobiology Hamburg (ZMNH), University Medical Center Hamburg-Eppendorf, Hamburg, Germany)
Prostate cancer relapse prediction is a challenging task within computational pathology as tissue preparation and digitalization is not standardized. The different protocols lead to domain shifts, against which a deep learning model must be robust and focus on biological information rather than variations in appearance. We address this challenge through the usage of vision foundation models...
Go to contribution page -
Thomas White (Deutsches Elektronen-Synchrotron DESY)Poster and Lightning Talk
The "Millepede" algorithm, for refining the geometry of complex multi-component particle physics detectors, was introduced in the late 1990s. The same approach, and in fact the same software, can be applied to serial crystallography for refining the positions and orientations of segmented X-ray detectors. The Millepede approach allows simultaneous joint refinement of the detector geometry...
Go to contribution page -
Niklas Laasch (Institute of Computational Neuroscience UKE Hamubrg)Poster and Lightning Talk
The mammalian cerebral cortex displays a complex network architecture supporting high-level cognition, yet the developmental rules driving its formation remain poorly understood. Current computational models often rely on static templates or global optimization, failing to capture the physical morphogenesis of neural tissue. Here, we present a multi-scale generative model integrating...
Go to contribution page -
Rudy Alexandro Garrido Veliz (Universität Hamburg)
In an ever-changing political and societal system, it has become a challenge for lecturers to effectively teach their courses on social studies (covering politics, society, and economy) with relevant and contemporary examples. This can lead to unprepared individuals in terms of media literacy, against problematic ideologies and propaganda.
Go to contribution page
A dashboard has been developed that aims to help... -
Thomas Asselborn (Universität zu Lübeck)Poster and Lightning Talk
Historical documents remain difficult to digitise accurately, as OCR systems struggle with niche fonts, paper degradation, physical damage, and handwritten annotations. Consequently, OCR results often contain errors that impair the usability of archives. We examine two machine learning-based approaches to OCR post-correction. The first uses the LLM Llama 3 to identify, correct, and reconstruct...
Go to contribution page -
Nils Leusmann (Universität Bremen / Data Science Center)Poster and Lightning Talk
Data management plans are as divers as the domains and data they describe. Utilizing the open source data-science tool EvoAl it is possible to introduce an structured way to represent research projects. Using model-driven software engineering EvoAl defines an domain specific language describing the data and experimental setup. This formal way allows users to quickly identify what the project...
Go to contribution page -
Marc Schulder (IDGS, Universität Hamburg)Poster and Lightning Talk
Resources and research on sign languages are sparse and can often be difficult to locate. Few centralised sources of information exist. This poster presents two repositories that aim to improve the findability of such information through the implementation of open science best practices. The sign-lang@LREC Anthology is a repository of publications on sign languages in the series of...
Go to contribution page -
Mr Hagen Peukert (Universität Hamburg)
This contribution presents collected requirements and architecture of an infrastructural data transfer software, DTF (Data Transfer Facilitator), using the methodological set up of ADD (Attribute-Driven Design). DTF facilitates the transfer of data from independent research programs to a sustainable data repository enriching the research data with metadata and transferring it to established...
Go to contribution page -
Amy Isard (IDGS, Universität Hamburg)
The research project “A Fresh Look. Visualising Digitised German-Jewish Archives” aims to investigate human information-seeking strategies in archival contexts that go beyond targeted, systematic searching, namely open-ended, curiosity-driven browsing based on the available visual information. We describe our work with two contrasting archives. First the Jerusalem Guestbook of Moshe and Miryam...
Go to contribution page -
Moldir Berkaliyeva (PhD Student)
Prenatal neurodevelopment research relies on 4D ultrasound imaging, yet manual frame-by-frame annotation of fetal body parts remains labor-intensive and inconsistent across human coders. This work adapts DeepLabCut for automated markerless fetal pose estimation across 12 anatomical bodyparts in 301 videos from 42 fetuses. We introduce a robust pipeline featuring fetus-level data splitting to...
Go to contribution page -
Heiko Bornholdt (Universität Hamburg)Poster and Lightning Talk
Modern research increasingly relies on distributed software, from AI workloads and simulations to data analysis across many machines. To cut latency, save bandwidth, and protect sensitive data, these applications run at the "network edge", near the data. But edge environments are messy: devices are scattered, diverse, and connected through untrusted networks, making deployment...
Go to contribution page -
Sebastian SĂĽnkler (HAW Hamburg)Poster and Lightning Talk
Evaluating commercial search engines is severely limited because providers offer no API access for collecting search results or AI summaries. Furthermore, independent data collection is limited by the dynamic nature of client-side rendering for these generative features. To overcome this, we present the updated Result Assessment Tool (RAT 2.0), an open-source Python toolkit comprising three...
Go to contribution page -
Mr Enrico Milutzki (Hub of Computing and Data Science)
As climate change intensifies flood risks, effective coordination of emergency forces becomes critical. This research explores the design of digital situation maps to support informed decision-making during urban flood disasters. By utilizing Design Science Research and expert insights, we developed a web-based prototype that enables structured, cross-organizational data input and...
Go to contribution page -
Subhankar Swain (Indian Institute of Technology, Kharagpur, India)
Memes, as a widely used mode of online communication, often serve as vehicles for spreading harmful content. However, limitations in data accessibility and the high costs of dataset curation hinder the development of robust meme moderation systems. To address this challenge, in this work, we introduce a first-of-its-kind dataset –TOXICTAGS consisting of 6,300 real-world meme-based posts...
Go to contribution page -
Duncan Paterson (Jinntec Gmbh)Poster and Lightning Talk
TEI Publisher 10 & eXist-db 7.0: New Horizons for XML Publishing
TEI Publisher 10 introduces Jinks, a visual application manager that enables editors to compose, generate, and upgrade digital edition applications without losing customizations — drastically lowering the barrier to creating and maintaining TEI-based publishing projects.
eXist-db 7.0.0-beta3 represents a...
Go to contribution page -
Mia Le (Bernhard Nocht Institute for Tropical Medicine)Poster and Lightning Talk
We developed ViMOP (Virus Metagenomics Outbreak Pipeline), a Nextflow pipeline based on our long-term experience in nanopore field sequencing in small stationary and mobile laboratories in Sub-Saharan Africa. ViMOP is designed to be robust and easy to use on a laptop for the automated analysis of nanopore reads generated directly from human clinical samples. The pipeline performs de novo...
Go to contribution page -
Johannes Keyser (Universität Hamburg)
For decades, sports and exercise research has advanced by isolating single variables or physiological signals to study behavior and performance. While informative, it has become increasingly clear that such reductionist methods struggle to capture the complex, dynamic, and context-dependent nature of sports and exercise behavior. Although advances in data analysis, artificial intelligence, and...
Go to contribution page -
Dr Willi Leinen (Helmut Schmidt University)Poster and Lightning Talk
Research software engineering for compute- and data-intensive software in high-performance computing (HPC) environments requires performance analysis, particularly insights into hardware utilization through low-level metrics. However, collecting and interpreting these metrics can be challenging and error-prone. We present xbat (extended benchmarking automation tool), an open-source,...
Go to contribution page -
Pierre Sfendules (Universität Hamburg), Mr Volker Boß (CAU KIel)Poster and Lightning Talk
We present a digital humanities project on “epochal biography” in nineteenth-century theological historiography. The project investigates works with title patterns such as “[Name] and his/her time” as a neglected genre between biography, church history, and cultural diagnosis. We are building a curated research database that combines bibliographic metadata, authority data, paratextual...
Go to contribution page
Choose timezone
Your profile timezone: