Jakob Nikolas Kather
jnkt.bsky.social
Jakob Nikolas Kather
@jnkt.bsky.social
Medical Doctor | Clinical Artificial Intelligence | Professor of Medicine and Computer Science at TU Dresden, Germany | X: jnkath | web: www.kather.ai
ChatGPT when I ask questions about breast cancer treatment 🙄
May 5, 2025 at 3:47 PM
A random photo which I posted on Google Maps years ago has 1.3M views now... approximately 1000x more than most of our scientific publications 😭
April 23, 2025 at 12:25 PM
happy weekend from our research group at @tudresden.bsky.social, credit to @zcarrero-katherlab.bsky.social
March 28, 2025 at 4:19 PM
Happy to share our latest paper in @ai.nejm.org, led by Isabella Wiest: "Deidentifying Medical Documents with Local, Privacy-Preserving Large Language Models: The LLM-Anonymizer" ai.nejm.org/doi/full/10....
You can use our open source tool with local LLMs to robustly de-identify medical documents.
March 27, 2025 at 6:32 PM
We have open Postdoc / PhD student positions in computational pathology and agentic AI, based in 🇩🇪🇪🇺 Ideally with a CS background, but we do 🧡 biologists with self-taught Python and ML skills
March 18, 2025 at 6:41 AM
It is happening. Pubmed (pubmed.gov) is gone. And so is cancer.gov. Scientists, what else is needed to wake up?
March 2, 2025 at 2:48 PM
Very happy to share our latest paper in @naturecomms.bsky.social, led by Kevin Boehm and Omar El Nahhas! Excellent collaboration between @mskcancercenter.bsky.social and @tudresden.bsky.social. We show in three large cohorts that multimodal AI improves breast cancer recurrence prediction.
March 2, 2025 at 2:34 PM
although they deny it...
February 1, 2025 at 11:41 AM
Very happy to share our latest paper, led by Jan Clusmann: Prompt injection attacks on vision language models in oncology. We show how VLMs in healthcare can be easily hacked and prompted to create harmful output. www.nature.com/articles/s41...
February 1, 2025 at 10:57 AM
Discussing science at University of Bern today ☺️🇨🇭🧀
January 31, 2025 at 1:44 PM
Very happy to share this „viewpoint“ article which was just published in @naturecancer.bsky.social ,
www.nature.com/articles/s43...
January 30, 2025 at 6:18 PM
Microsoft copilot is great - I know have useless links to a broken chatbot hovering all over my Powerpoint presentation elements 👍
January 28, 2025 at 8:56 PM
I 🧡 AI hallucinations… if you transcribe an empty audio file with Whisper, this is what you get:
January 23, 2025 at 7:42 AM
I’ve been an examiner for PhD defenses in Germany, UK, Finland, and the Netherlands. NL has
- robes
- Latin (!) sermon
- real medieval weapons on display
ChatGPT/DALLE gets the vibes quite well in this image 😃 (to be fair, the Finland experience is great too)
January 23, 2025 at 5:43 AM
ASCII art is the greates achievement of modern LLMs
January 22, 2025 at 7:54 AM
Just a reminder that this was the state of the art 2022 - we've come a long way in medical applications of generative AI. Source: www.nature.com/articles/s41...
January 22, 2025 at 7:32 AM
Very happy to share our new paper on AI-based biomarkers 🤩 AI can support clinical decisions in oncology. We introduce a clear terminology and provide key example for experimental and mature AI systems in oncology decision making. Led by Marta Ligero.
www.sciencedirect.com/science/arti...
January 15, 2025 at 9:12 PM
very happy to share our latest preprint, led by Michaela Unger: www.biorxiv.org/content/10.1... We are using Deep Learning to improve the detection of biomarkers in cancer genomic data 👇 This is an emerging field, much more to come.
January 9, 2025 at 11:25 AM
We need to make it easier to create pan-European startups 🇪🇺. Happy to support the EU Inc. initiative -> www.eu-inc.org
December 31, 2024 at 9:30 AM
Very happy to share our article about AI agents in cancer research and oncology, with Yongju Lee, Dyke Ferber, Jennifer E. Rood, & Aviv Regev, out in Nature Cancer now -> www.nature.com/articles/s43...
December 18, 2024 at 9:13 AM
Hey science bluesky: The term "computer-aided diagnosis" is on the decline. My hypothesis is that it is being replaced with Deep Learning / AI. Classical rule-based computer-aided diagnosis systems never really worked in practice. Is this view too simple?
December 11, 2024 at 1:55 PM
Continuing the rant against the "editorialmanager" system used by Elsevier. You have to respect that it would be difficult to come up with a *worse* user interface than this one. The only truly worse pieces of software out there are the electronic health records systems used in our hospitals.
December 11, 2024 at 7:19 AM
The world of medicine, and many businesses related to it, should brace for massive changes by LLMs - the cost (and value) of knowledge and reasoning in healthcare is rapidly going down
December 6, 2024 at 10:36 AM
another example of the user interface of editorialmanager. Who came up with this? We academics deserve better...
December 5, 2024 at 8:40 PM
I am happy to peer-review articles for scientific journals for free. However, I am not motivated to complete repetitive drop-down menus at every stage of the revision process. Even when managing manuscripts as an editor, I prefer semi-structured reviews over standardized drop-down menu selections.
November 30, 2024 at 10:12 AM