Stanford’s Dr. Sylvia Plevritis is pioneering a new approach to cancer research using artificial intelligence to decode the spatial architecture of tumors—paving the way for personalized therapies and future breakthroughs in cancer treatment.

In a world where cancer remains one of the most complex and elusive diseases, a Stanford scientist is helping to redefine how we understand—and potentially conquer—it. Dr. Sylvia Plevritis, a professor at Stanford University and a pioneer in biomedical data science, is advancing a new frontier in cancer research by combining artificial intelligence with tumor biology.

Her work centers around what she refers to as the “cellular neighborhood” inside tumors—a concept that goes beyond the genetic profile of a cancer cell. Instead of viewing cancer as a mere collection of rogue cells, Dr. Plevritis and her team are mapping how diverse cell types interact within a tumor, forming what she calls the colocateome.

“Tumors aren’t just a mass of bad cells,” she explained during a recent Science Friday interview on NPR (June 6, 2025). “They are ecosystems, with immune cells, connective tissues, and signaling patterns all communicating. Understanding those relationships is key to knowing why some treatments work and others don’t.”

This pioneering concept reframes the tumor as a complex, spatially organized ecosystem—one in which immune cells, stromal components, and cancer cells form structured interactions that influence how a tumor behaves and responds to therapy. Through her lab at Stanford’s Center for Cancer Systems Biology (CCSB), Dr. Plevritis is developing tools to decode these structures using advanced machine learning and image analysis technologies.

With the help of artificial intelligence, her team analyzes massive datasets—ranging from pathology slides and spatial transcriptomics to molecular assays and electronic health records. These data are computationally modeled to create detailed spatial maps that reveal where different cell types “colocate” within tumor tissue and how they behave in proximity to one another.

This multidimensional view enables predictions about which therapies are likely to be most effective for specific patients. For example, the presence or absence of immune cells at the invasive edge of a tumor can serve as a biomarker for responsiveness to immunotherapy. AI models developed in her lab help quantify these spatial arrangements with precision far beyond human capability.

“We’re seeing that AI can help us interpret the high-dimensional relationships in tissue architecture—relationships that are not always apparent even to the trained human eye,” Dr. Plevritis noted. “This can guide oncologists in selecting more targeted therapies.”

Her research, published in journals such as Nature Communications, has demonstrated that spatial signatures derived from the colocateome can offer more predictive power than genetic profiling alone. For instance, in a 2021 study (Zhao et al.), her team showed how spatial clustering of immune and tumor cells in breast cancer samples could predict patient outcomes and therapeutic response.

While the promise is exciting, the science is intricate. Mapping these micro-environments requires not only advanced imaging technologies—like multiplexed immunofluorescence and spatial transcriptomics—but also the computational infrastructure to process and interpret petabytes of data. Fortunately, modern AI tools are rising to meet this challenge.

As cancer care continues to evolve, the integration of artificial intelligence into diagnostics and treatment planning is emerging as a game-changer. Rather than replacing oncologists, AI supports them—providing faster insights, richer tumor maps, and ultimately, more personalized care strategies.

Dr. Plevritis’s work exemplifies the synergy between biomedical science and cutting-edge computational models. By visualizing the tumor as a spatially complex community, AI enables researchers to not only map the cancer landscape but to simulate how it will change under different therapies—potentially unlocking new frontiers in drug development and precision oncology.

At the intersection of data and biology, this is the kind of innovation that could mark a turning point in the fight against cancer.

And perhaps one day, thanks to advances in machine learning, pattern recognition, and molecular modeling, artificial intelligence won’t just help diagnose and manage cancer—it may help cure it. With each dataset, each tumor map, and each model refined by AI, we move one step closer to a future where cancer is not a mystery, but a solvable equation.

To support ongoing research and patient care, consider donating to the American Cancer Society. Your contribution fuels breakthroughs like these and brings hope to millions of families around the world.
https://donate.cancer.org

Citations & Resources:

• Sylvia K. Plevritis, PhD – Stanford Profile

• NPR Science Friday (June 6, 2025): Mapping Cancer’s Hidden Architecture – Listen Here

• Zhao, E., Stone, M. R., Ren, X. et al. “Spatial transcriptomics at subspot resolution with BayesSpace.” Nat Commun 12, 1900 (2021). https://doi.org/10.1038/s41467-021-21947-1

• Center for Cancer Systems Biology (CCSB) – ccsb.stanford.edu

• American Cancer Society – https://donate.cancer.org

• You might enjoy listening to AI World Deep Dive Podcast: