Krauthammer Lab maps SSc disease subtypes
In npj Digital Medicine, DQBM’s Krauthammer Lab and EUSTAR collaborators present a generative temporal framework that learns organ-specific latent processes, imputes missing data, and clusters patient trajectories. The model identifies a high-risk subgroup with limited skin disease but elevated lung and heart complications, refining risk stratification beyond cutaneous subsets.
Trained on 14,000 patients and 67,000 visits, the approach outperforms baselines and supports a prototype clinical decision tool for early severity prediction.
DOI: https://doi.org/10.1038/s41746-025-01962-y.