The Dumitrascu Group develops interpretable machine learning models that integrate gene expression, morphology, and mechanics to uncover how cells organize, communicate, and regenerate within complex tissues. By modeling the shared principles between development, regeneration, and wound healing, the group seeks to understand the common rules of biological self-organization that enable tissues to build, repair, and renew themselves. Current directions in the Dumitrascu Lab explore:

• Mathematical modeling of morphogenetic and metabolic dynamics
• Mechano-transcriptomic modeling and multi-scale causal inference
• Contrastive graph learning for tissue niche discovery
• Counterfactual inference in regenerative and developmental systems