Matthias A. Fink

GPT-4 enabled near-perfect creation of pancreatic ductal adenocarcinoma synoptic reports from original reports and accurate resectability categorization using chain-of-thought prompting. @imagingtoronto @RajeshBhayana_ @SatheeshToronto bit.ly/4cqQ8I2

Performance of an Open-Source Large Language Model in Extracting Information from Free-Text Radiology Reports doi.org/10.1148/ryai.2… @BleGuellec @CHU_Lille @PasteurLille #Vicuna #reporting #ML

In this targeted editorial, Dr. @_rahsepar of @NURadiology discusses the ability of LLMs to simplify radiology report impressions, beneficial for patients and patient-provider communication. doi.org/10.1148/radiol…

Measure what matters: Encourage researchers to track & report #SustainabilityMetrics alongside their AI research. #Transparency is key! #RadAIchat #AIforGood #RadAIchat

@pintodrad nice! 🤩

This is amazing!! 🤩🤩🤩

@pintodrad @radiology_rsna @tmdratsch @UKKoeln @EuSoMII wow!! congrats 🥳👏👏👏

Wow!!! What a nice Our paper on #automationbias in #mammography made it to the top10 most cited in 2023 in @radiology_rsna!!! 🥳🥳🥳🥳 Amazing work with @tmdratsch (@UKKoeln) and the rest of the team!! @EuSoMII pubs.rsna.org/doi/10.1148/ra…

Fascinating discussion led by @MatthiasAFink, moderated by Dr. Jordan Perchik from @UABRadiology, exploring the effectiveness of LLM ChatGPT and GPT-4 in mining CT reports for lung cancer. Impressive performance by GPT-4, outshining ChatGPT! Is a subscription worth it? #RSNA23

Newly-published study investigates the impact of structured reporting on the linguistic standardization of #radiology reports. (@Voss_MD et al.) #EuropeanRadiology 🔗 buff.ly/3Ew8IQo

#ChatGPT and #GPT4 may both be useful for analyzing free-text notes and extracting insights, study finds. #AIinHealthcare #AI @OpenAI #LungCancer ow.ly/FZML50POIET

@barrygoldman1 @francesarnold @EvolvedBiofilm @kalpenn Don't take it too seriously. #TBBT is a great show that also tackles some scientific issues in an ironic and colorful way. I first saw @francesarnold on #TBBT and became an even bigger fan of her after her amazing talk (where she showed this same group photo) @lindaunobel 😍

@francesarnold @EvolvedBiofilm @kalpenn i could not stomach it. and would guess it did more harm than good. show me wrong. the public perception of science is the pits!

Remembering my fun day at #BBT ... #science doesn't get better than this. Biggest accolade of my life, except for a nice party in Stockholm, of course.

💡Excited to announce a cross-discipline, worldwide collaboration creating common shared guidelines for accountable and ethical use of #ChatGPT and #LLMs in #science ✅Introducing the @CANGARU_check guidelines project. Find our correspondence in @Nature 🎓Unified #Reporting is critical to avoid conflicting guidelines 🌍Together with AI guidelines developers, journal editors, publishers, scientists, researchers, and representatives of regulatory agencies and others, we can shape responsible reporting, disclosure & use of GPTs/LLMs to protect and preserve the research 2.0 #Standardization #guidelines #ArtificialIntelligence 🌍 We're planning a global #DelphiConsensus, orchestrating a chorus of diverse voices to ensure everyone is heard #inclusivness 👉 doi.org/10.1038/d41586… 👉go.nature.com/43dxnwe @GSCollins @EQUATORNetwork @C0PE @USC_Urology @EASEeditors @TheLancet @APSphysics @LancetDigitalH @eLife @EricTopol #WAME @NatureMedicine @NatMachIntell @PNASNews @JAMA_current @J_A_C_S @ISMTE @ConnerGanjavi @MichaelEppler7 @yanshan_wang @MarieFSouliere

#LLM #AI #AIrounds - Nature article on LLM encoding clinical knowledge nature.com/articles/s4158…

The panel discussion on "Artificial Intelligence and Medicine" was undoubtedly one of the many highlights from #LINO23. Together with all of this year's sessions, a recording can be found in our Mediatheque: ow.ly/KmQg50P62U2

The anatomy in chest radiographs (CXR) is typically difficult segment manually due to the ambiguity introduced by overlapping structures. We Adress this in our paper and publish models to segment 157 anatomical labels in CXR: arxiv.org/abs/2306.03934

well deserved 👏👏👏

What a pleasure to be part of such a network of like-minded academic radiology trainees with a passion for research and radiology journalism @DRG_de @FJR_FFZ_DRG

@Radiology_AI @kalpathy1 @jonherochung @DukeRadiology Happy Valentine's Day!

Fully automated pipeline assesses muscle and fat tissue on routine chest CT doi.org/10.1148/ryai.2… @kalpathy1 @jonherochung @DukeRadiology #DeepLearning #BodyComposition #MachineLearning

@Kaggle grandmaster @IanPanMD details his solution to the #RSNA20 Pulmonary Embolism Detection challenge doi.org/10.1148/ryai.2… @RSNA @thoracicRad @BWHRadEdu #PulmonaryEmbolism #AI #ML