Within the Horizon Europe Project HEREDITARY (running from 2024-2027), we collaborate with different partners from medical research, machine learning, databases and other fields on the visualization of multimodal, distributed clinical and research data, building various visualisation prototypes and demos. Many of the demos are developed as part of the Work Package 5 (WP5) efforts, which focusses on visualisation. This webpage demonstrates many of the visualization approaches we are developing, making them available for interested parties to try out.
Clusters in Focus
Exploring tabular datasets to understand how different feature pairs partition data into meaningful cohorts is crucial in domains such as biomarker discovery, yet comparing clusters across multiple feature pair projections is challenging. We introduce Clusters in Focus, an interactive visual analytics dashboard designed to address this gap.
Neurodegen-Vis
Healthcare records, measurements and datasets can quickly become highly complex in terms of storage, anonymisation and interpretation. With Neurodegen-Vis, we present a practical toolkit to anonymise, visualise and help interpret medical data in a Visual Analytics dashboard, with seamless integration of a Large Language Model (LLM).
OnSET
OnSET is designed to enable a more explorative ontology querying experience for non-expert users, and to build a quick understanding of the scale and nature of a KG that is in use. The toolkit consists of multiple smaller systems, complimenting each other.
ParamInter
ParamInter is a novel toolset to explore continous input-output parameter sets interactively through guidance by XAI and UCQ. The approach interpolates between selected input samples over an embedded subspace, guiding the user towards required solutions for their target output parameters.
Gut-Brain Viewer
The Gut-Brain Viewer visualizes gut microbiota-brain links from LICA outputs via three linked views. An interactive microbiota matrix shows organism Z-scores per component with thresholding and selection. A modality stacked-bar view displays percent contributions and loads the selected modality. Interactive brain slices and volumetric renders show voxel Z-scores for SN, ECN, or DMN from 4D NIfTI files.
Multi-Component State Model Visualisation
This component is concerned with the visualisation of a multi-component state model coupled (biological) system. Each component can be modelled through its own individual state model, including dependencies on the environment such as ion concentrations, voltages or other variables.
Differential Cell Signaling
In medical applications, the increasing availability of single-cell RNA sequencing data from pathological and healthy samples has led to the development of methods for differential analyses, i.e. extensions of classic bioinformatics analyses which identify significant biological alterations in the pathological condition with respect to an healthy reference. A first draft of the dashboard shows a preview of the graph.
Alvis
Amyotrophic Lateral Sclerosis (ALS) is a fatal neurodegenerative disease affecting over 300,000 people worldwide. Currently, there is no cure—only treatments aimed at improving patients’ quality of life. We introduce ALviS, a prototype designed for analyzing ALS datasets, providing a foundational framework for visualizing and exploring the multivariate ALS dataset.
Neuronal Brain Simulation
The Brain Connectome and Activity Data dashboard visualises a simulation of the brain. On the left, we have a shared view between the neuron graph and a perspective view. On the right, the varying quantities described above are displayed, for one selected neuron, as a coloured line chart. The colour intensity is used to indicate areas of high activity, although that can be adjusted.
Talk To Your Gene-Disease Graph
By enabling users to communicate in natural language rather than technical query languages like SPARQL, Talk To Your Graph (TTYG) reduces technical barriers for non-expert users. TTYG enables intuitive access to complex data while remaining independent of the underlying schema and resilient to subsequent modifications in the knowledge graph’s organization. The Talk To Your Gene-Disease Graph demonstrator is able to answer natural language questions on a custom-curated knowledge graph about gene-disease associations. It allows for identification of novel biological targets of new potential drug agents and for understanding of the disease biology - related biological processes, mechanism, tissue/organ/cell profiling, interacting partners, etc.
Visual Inspection of EEG Data Frame and Model
EEG measurements are an important biosignal used to assess brain health. For example, EEG helps diagnose neurodegenerative diseases and monitor patient status. Selecting which EEG signals to use for model training is time-consuming and often uncertain. Visualization tools can help screen EEG time windows and make selection decisions faster. A dashboard should display OK/NOK labels and allow comparison against clusters, frequency bands, and sensors.
