How One Moffitt Team Is Using AI To Help Decode Cancer Care
Key Takeaways:
• Moffitt researchers developed ECHO, an AI framework designed to organize and interpret complex cancer data
• ECHO shows the reasoning behind its conclusions to support transparency and human oversight in cancer research
• The team hopes ECHO will help personalize multiple myeloma treatment by identifying why therapies succeed or fail
At first glance, AI in cancer care can sound intimidating, even impersonal. But for researchers Ariosto Silva, PhD, Kenneth Shain, MD, and their team at Moffitt Cancer Center, AI is not about replacing physicians or making mysterious decisions behind a black box. It is about helping researchers and clinicians connect the dots faster so they can ask better questions, uncover deeper insights and ultimately improve patient care.
That vision is at the center of the team’s research being presented at the 2026 American Society of Clinical Oncology Annual Meeting.
ECHO is an AI-powered framework designed to help oncology researchers and physicians organize and interpret the massive amounts of data involved in cancer research and patient care. Instead of simply producing answers, ECHO shows users the reasoning and steps behind its conclusions, creating a more transparent and collaborative approach to AI in biomedical research and medicine.
“ECHO is a tool for team science,” Silva said. “It replaces neither the thinking nor the science. It helps researchers connect data, samples and clinical information so they can focus on solving the why.”
Turning Mountains of Data Into Meaningful Insights
Modern cancer research generates enormous amounts of information. Researchers may have access to tumor samples, genomic sequencing, lab results, imaging data, treatment histories and clinical outcomes, but these pieces often live in separate systems and formats.
According to Silva, researchers can spend months simply trying to locate, organize and connect those datasets before meaningful analysis even begins.
ECHO streamlines that process.
Much like a conversational AI tool, users can ask ECHO a research question in plain language. The system then breaks the problem into smaller steps, identifies the relevant databases and software tools, retrieves the information and produces a detailed report. Unlike traditional GenAI systems that simply provide an answer, ECHO also shows the full process it used to arrive there.
That transparency is one of the project’s defining features.
“The answer alone is not enough,” Silva said. “You need to understand how the system reached that answer and whether the reasoning makes sense.”
Supporting Personalized Cancer Care
The team’s work is especially focused on multiple myeloma, a complex blood cancer that often requires combinations of therapies over many years. While treatments can place the disease into remission, many patients eventually relapse and require new therapies.
“One of the biggest challenges in oncology is determining which treatment is most likely to work for an individual patient and why,” Shain said.
Today, many treatment decisions rely on population-level clinical trial data. ECHO aims to help researchers move closer to truly personalized care by integrating biological, clinical and molecular data into a more complete picture of each patient’s disease.
Shain explains that ECHO is not designed to function as a “magic eight ball” that simply predicts outcomes. Instead, it is meant to uncover the biological reasoning behind treatment responses and resistance.
“If a therapy may not work, physicians need to understand why,” Shain said. “That understanding can help researchers discover new strategies or identify alternative treatments.”
Keeping Human Expertise at the Center
As AI becomes more common in everyday life, concerns about inaccurate outputs and AI hallucinations continue to grow. Silva says those concerns are exactly why transparency and human oversight were built into ECHO from the beginning.
“AI should help us understand cancer, not replace physicians or scientists,” Silva said. “The expertise still comes from the people interpreting the information.”
The approach reflects what Silva calls “human-in-the-loop” science, where AI accelerates the work, but experts remain responsible for evaluating the findings and making decisions.
