Latent Patient Representations using Single-Cell Transcriptomics (RUST)
This project aims to develop an AI-based computational tool that utilizes advanced machine learning techniques to generate patient-specific 'embeddings'—compact representations from vast amounts of cell data, which reveal unique disease signatures. Compared to sparse, high-volume raw sequencing data, the idea behind these embeddings is to compress disease relevant information into a lower dimensionality format for broad usability. By analyzing approximately 2 million cells from 200 patient samples, we seek to create a framework that follows the FAIR principles for data management, enhancing the accessibility and reusability of clinical research data.
Combining medical expertise, bioinformatics, and machine learning, the project addresses the challenges of integrating and analyzing vast amounts of biomedical data to capture patient-specific disease signatures. This interdisciplinary approach is necessary to develop new computational models that extend beyond traditional cell-level analyses, providing deeper insights into patient-level disease states. We expect the project results to facilitate the comparison of disease conditions across patients, potentially leading to the identification of novel biomarkers and therapeutic targets.
Collaboration and Funding: This research is an exploratory effort, part of a broader efforts at RWTH Aachen to enhance data-driven medicine. Supported by an interdisciplinary team and international collaboration, the project embodies the intersection of computational life sciences and clinical application.
 |  |
Involved PIs
- Dr. Sikander Hayat
Translational Data Science, RWTH Aachen University Hospital - Prof. Rafael Kramann
Department of Medicine II, RWTH Aachen University - Dr. Jan Pennekamp
COMSYS
Our Team
Funding
SFFAIR002. Funded by the Excellence Strategy of the German federal and state governments.