AstroPath, celestial mapping applied to cancer, joins Anthropic's AI for Science: AI reaches oncology imaging

By Anthropic News (via OncoDaily) · October 31, 2026.

The Mark Foundation for Cancer Research (MFCR) has announced on X that AstroPath, the platform developed by Janis Taube and Alexander Szalay, has been selected for Anthropic's AI for Science program. Taube and Szalay are part of the Mark Foundation Center for Advanced Genomics and Imaging at the Johns Hopkins Sidney Kimmel Cancer Center.

According to the MFCR itself, AstroPath uses sky-mapping tools to image cancer. The source also says that AI for Science is now expanding to foundations and research institutions. That is all the material confirms. It does not say what resources, credits or specific support the project will receive, how Claude will be used, or what timelines or results are expected.

That is why it is best not to read too much into it. Being selected for a program is not a scientific result, and there are no figures here on diagnostic accuracy or clinical impact.

Our take: what makes this interesting is the approach. Techniques used to map the sky (large images, millions of objects, layered datasets cross-referenced against each other) are a good fit for tumor tissue, where every cell is a point to be located and classified. It is a recurring pattern: methods from one big-data discipline migrate to another, with AI acting as both accelerator and interpreter.

The broader shift matters too. If the program opens up to foundations and institutions, frontier AI stops being the preserve of labs and companies and enters biomedical research in a structured way. That is a good sign for the long-term thesis: the road to eradicating disease runs less through a single big breakthrough and more through many tools that improve how biological data is read.

An honest caveat: in oncology, the gap between good image analysis and a change in how patients are treated is wide and requires clinical validation. We will have to wait for published results to know how much this collaboration really contributes.