Lorin Crawford Team Publishes 2026 AI Cancer Study, Challenging Bigger-Data Assumptions
Updated
Updated · Microsoft · Jun 9
Lorin Crawford Team Publishes 2026 AI Cancer Study, Challenging Bigger-Data Assumptions
1 articles · Updated · Microsoft · Jun 9
Summary
A Nature Methods study published Tuesday shows Lorin Crawford’s team can use AI to read how individual cancer cells behave and interact with their surroundings, aiming to explain why similar tumors can respond differently to the same drug.
Project Ex Vivo focuses on “cell state” rather than mutations alone, arguing that lab-grown models often miss key tumor behaviors seen in patients and can mislead drug testing and treatment selection.
The researchers found AI learns more from exposure to a wider range of cell behaviors than from simply adding more data, undercutting a common belief that scale alone will solve cancer-modeling limits.
Microsoft, the Broad Institute and Dana-Farber say the approach could improve patient matching for existing therapies and trials, and eventually help develop drugs that target or even shift a tumor’s state.
Started in 2022, the effort now uses virtual experiments to narrow lab work, with the next step to define and validate cell states across cancers so doctors can use them in treatment decisions.
Can AI's focus on cancer 'behavior' finally outsmart a tumor's ability to evolve and resist our best drugs?
What new class of drugs is needed to target the elusive 'cell states' that AI can now identify?
If AI prioritizes data diversity over volume, does this upend the 'big data' strategy in medical research?
Lorin Crawford’s 2026 AI Cancer Breakthroughs: Generative Models, Data Diversity, and the End of the “Bigger is Better” Era
Overview
Lorin Crawford is leading a major shift in cancer research by applying generative AI to tackle the complexity of the disease. With a strong background in statistical genetics and machine learning, he challenges the traditional view that more data always leads to better AI models. Crawford’s work, highlighted in a 2026 Cell review, shows that generative AI can recognize complex patterns and synthesize diverse data, making it essential for new discoveries and better cancer care. His approach moves beyond reductionist frameworks, focusing on holistic, dynamic models that improve diagnosis and treatment strategies in oncology.