Georg Wölflein

@katherlab @univofstandrews

[Georg Wölflein @gwoelflein]

🚀 New preprint! 🚀

LLM Agents Making Their Own Tools? Meet ToolMaker 🛠️

Our latest work explores how LLM agents can autonomously transform scientific code repositories into LLM-compatible tools – paving the way towards fully autonomous scientific workflows! 🧑‍🔬🤖

Paper: https://t.co/mjLUQ1jqi2
Code & benchmark: https://t.co/cP2LpG1OkJ

Special shoutout to @Dykex6 for proposing the idea and being an incredible collaborator! 🙌 Also grateful to @DanielTruhn, @Oggie_A, and @jnkath for making this happen!

#AI #LLMAgents #ToolCreation #AI4Science #MachineLearning #ComputationalPathology #ToolMaker

[Dyke Ferber @Dykex6]

Very happy to share our recent article in Nature Cancer on AI Agents for Decision Making in Cancer :)

@jnkath

[Ognjen (Oggie) Arandjelović @Oggie_A]

Congrats to my #PhD student, @gwoelflein, for having his paper "LLM Agents Making Agent Tools" accepted to ACL (@aclmeeting), the top #conference on comp #linguistics. It's a very cool paper - do read it! https://t.co/ZCF24eKZH1

#AI #LLMs #AgenticAI #ArtificialIntelligence

[Jakob Nikolas Kather @jnkath]

We are looking to hire a PostDoc at @katherlab to develop AI multimodal foundation models for uncovering cancer metastasis mechanisms and identifying targetable biomarkers for patient treatment. 🩻🔬🧬
If you’re interested, please apply here: https://t.co/lqKjpUeYPR @tudresden_de

[Jakob Nikolas Kather @jnkath]

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

[Jack Morris @jxmnop]

people underestimate the mental cost of outsourcing code to Copilot/Cursor

it's a mortgage: quick progress now at the expense of not understanding your own codebase

it may be that beyond simple line autocomplete, it's more efficient in the long run to do everything yourself

[Jakob Nikolas Kather @jnkath]

So proud of our new preprint "toolmaker" by @gwoelflein et al.: AI Agents building AI agents.

[Georg Wölflein @gwoelflein]

@science_wallet Thanks for the shoutout!

[Thomas Walle @science_wallet]

You can now use our single cell knowledge base

Cytopus 🐙
https://t.co/npiX7ZrLjl

with AI thanks to this neat tool:

[Jakob Nikolas Kather @jnkath]

Looking for Postdocs and Research software engineers 😊! Are you an AI researcher (CV / NLP / agents) with a fascination for oncology? Please reach out! We are based in 🇪🇺. Salaries are ok, cost of living is low, and we don't live in a dystopian totalitarian system 🤞

[Jakob Nikolas Kather @jnkath]

Happy to share our new paper from @katherlab , led by @Dykex6 , which was just published in @NatureComms. In-context learning enables general-purpose vision language models to perform well on medical image analysis tasks. We don't need to train a neural network at all - just provide a handful of examples at inference time.
https://t.co/Lz6VwRoNmZ
PS: follow me on https://t.co/TS8XGbyf39

[Jakob Nikolas Kather @jnkath]

Finally out in @NatureProtocols : our group's workflows for end to end computational pathology. Compatible with UNI and other foundation models. Led by @ElNahhasOSM from @katherlab
journal link: https://t.co/oa2HFvH4Kg
full text: https://t.co/8zKLRNAgjB

[Jakob Nikolas Kather @jnkath]

New preprint from @katherlab: large language models for end-to-end clincial trial matching. Our system parses study protocols and patient records, and matches them. We evaluate thoroughly against human experts and find >90% correspondence. Led by @Dykex6 🤩 https://t.co/vTnTVd19gq
@NCT_UCC_DD @NCT_HD @DKFZ @DataScienceDKFZ @uniklinik_hd @myESMO CC @tompowles1 @NEJM_AI @Agrawal_Mri @alexia_zz @PepPamies

[Georg Wölflein @gwoelflein]

🧬 Seemingly every few weeks there’s a new pathology feature extractor / foundation model! 🚀

To keep up, I compiled a comprehensive list of such models with key details like SSL pretraining methods, architectures, dataset sizes, and more.

Check it out: https://t.co/10NLG3fyVJ

[Jakob Nikolas Kather @jnkath]

Finally out in @NEJM_AI 🤖, led by @Dykex6 from @katherlab: "GPT-4 for Information Retrieval and Comparison of Medical Oncology Guidelines". We've developed a RAG pipeline using the GPT-4 API to answer medical oncology questions based on guidelines. Previous studies have claimed that "large language models (LLMs) are not suited for oncology" -- but our data debunk these claims! Using RAG, LLMs are very good at medical decision-making in oncology -- as we show in a quantitative evaluation. There is still room for improvement, but LLMs+RAG are the way to go ✨

Paper link: https://t.co/fw0j8OEBg3
Free-access link: https://t.co/GQQjbNCtId

@myESMO @ASCO @AACR @uniklinik_hd @Medizin_TUD @NCT_UCC_DD @NCT_HD @DataScienceDKFZ

[Georg Wölflein @gwoelflein]

@eccvconf Is there a template for the rebuttal?

[Jakob Nikolas Kather @jnkath]

Happy to share our latest preprint led by @ElNahhasOSM 🌟: A manual for our deep learning pipeline for pathology image analysis, tried and tested in dozens of projects, fully open source. @tudresden_de @nct_hd @NCT_UCC_DD @Medizin_TUD https://t.co/1qBCGb2Kdm

[KATY @project_katy]

Mark your calendars for @wacv_official 2023 in Waikoloa, Hawaii - #KATY partner @univofstandrews will present a paper there.

Check out the preprint: https://t.co/yW38o3PKRa
More information on the event: https://t.co/1SWyISX2uL

#AI #fightcancer #TCELL