Seema Saharan is a postdoctoral scholar in the Department of Radiology and Biomedical Imaging at UCSF, specializing in radiology image analysis, large language models (LLMs), and multimodal data integration. Her research combines deep learning, causal inference, and agentic AI frameworks to advance diagnostic accuracy, outcome prediction, and equitable implementation of precision medicine.

Her expertise spans data science, biostatistics, and AI-driven methodologies, with a focus on integrating radiology imaging, biomedical signals, and high-dimensional biomolecular data. She develops AI-powered diagnostic tools that leverage computer vision, neural networks, and multimodal fusion techniques for early disease detection, risk stratification, and personalized treatment strategies.

Seema also contributes to NIH-funded projects addressing Alzheimer’s disease, chronic pain, and disparities in molecular diagnostics access. Her work involves scalable pipelines for large claims datasets, NLP-driven extraction of unstructured EHR data, and transformer-based approaches (BERT, BioBERT, ClinicalBERT) for analyzing ctDNA testing pathways.

She holds a Ph.D. in Statistics with a Data Science Algorithmic focus, where she optimized statistical models of cytokine cascades transported by HDL/Plasma to inform cardiovascular and Alzheimer’s disease research. In collaboration with UCSF’s Cardiovascular Research Institute, she has developed LLM-enabled diagnostic frameworks that integrate cytokine biomarkers, clinical data, and biomedical literature, incorporating retrieval-augmented generation (RAG), SHAP-based explainability, and causal inference approaches.

Beyond research, Seema is passionate about building standardized AI ecosystems for healthcare that are interpretable, secure, and clinically impactful. She is also an experienced educator, serving as a lecturer at UC Berkeley Extension and California State University, East Bay, where she teaches courses in data science, AI, and biostatistics.
.