IWR News New IWR Member: Dr. Jovan Tanevski

My research group develops explainable machine learning and optimization-based frameworks for integrating imaging, single-cell, and spatial multiomics data to uncover the spatial organization and dynamics of tissues at molecular resolution. By reconstructing these organizational patterns across molecular layers, we aim to reveal mechanistic principles governing tissue organization in health and disease.

We ask whether the spatial organization of tissue carries meaningful signal, whether this signal is legible across molecular layers and data modalities, and whether spatial organization can predict clinical outcomes in an explainable way. Ultimately, we aim to build a mechanistic and predictive understanding of how molecular perturbations drive tissue reorganization and clinical outcomes.

To address these questions, we specialize in unsupervised relationship-based representation learning to identify tissue patterns and spatial dependencies that define the local and global organization of tissue. We also develop optimization tools for multiomics integration and spatiotemporal modeling, enabling the transfer of information across modalities and the reconstruction of tissue dynamics over time. Finally, we apply supervised learning from cross-conditional and longitudinal data to link tissue architecture and molecular patterns to clinical outcomes through graph neural networks.