Jakob Nikolas Kather, Professor of Clinical Artificial Intelligence at Dresden University of Technology, shared a post on LinkedIn:
“Our review on AI agents in cancer research and oncology is out in Nature Portfolio Nature Reviews Cancer.
Over the past few years, large language models have evolved from chatbots to something more interesting: agents that can plan, reason, and actually do things. We tried to make sense of what this means for cancer research and clinical oncology.
The basic idea is that AI is moving from classifying and predicting to orchestrating entire workflows. Scanning literature, analyzing data, proposing hypotheses, designing experiments. In the clinic, these systems could help piece together patient information across imaging, genomics, and clinical history.
We also talk about the unsolved problems: validation, regulation, workflow integration, automation bias (check out Danielle Bitterman’s recent work on this).
Of course, this field moves so fast that parts of this review are already outdated, but we hope it is still useful as a way to think about where things are heading.
Work led by Daniel Truhn, thanks to Shekoofeh Azizi, James Zou, Leonor Cerdá-Alberich and Faisal Mahmood for writing this together.”
Title: Artificial intelligence agents in cancer research and oncology
Authors: Daniel Truhn, Shekoofeh Azizi, James Zou, Leonor Cerda-Alberich, Faisal Mahmood, Jakob Nikolas Kather
You can read the full article in Nature Reviews Cancer.
More posts featuring Jakob Nikolas Kather.