_hylandSL - not here

machine learning for healthcare - principal researcher at microsoft research in cambridge UK 🌻 she/her
@hylandsl.bsky.social

[_hylandSL - not here @_hylandSL]

Trying this out, I miss using twitter for academic purposes: https://t.co/htCc1UBoko

[_hylandSL - not here @_hylandSL]

if your immediate thought was to ask "what exactly do you mean by biomedical discovery?" you are double-invited to apply, because we care about details

[_hylandSL - not here @_hylandSL]

if you find yourself wanting to understand how and why models work, in a way that could be useful for biomedical discovery, come to Cambridge (UK) for a postdoc at MSR: https://t.co/30ZdOVSi4A

[_hylandSL - not here @_hylandSL]

Our MAIRA-2 model for (grounded) radiology report generation from CXRs is out now! https://t.co/1vHcB6vXFt

[_hylandSL - not here @_hylandSL]

By open-sourcing RadFact, we hope others can use it, improve on it, and ultimately advance state of the art in automated radiology report generation!

https://t.co/5wbhdgbPUy

[_hylandSL - not here @_hylandSL]

my team's radiology report generation metric is now open source:

microsoft/RadFact: A metric suite leveraging the logical inference capabilities of LLMs, for radiology report generation both with and without grounding (https://t.co/HfvVKhXuEp)#MoreMetrics
https://t.co/O8BUiFKuGN

[_hylandSL - not here @_hylandSL]

We needed a new metric to support grounded report evaluation, but it was already evident when we were doing MAIRA-1 (https://t.co/V3HbiaBXuB) that even for text-only evaluation, existing metrics miss things we care about. RadFact targets correctness and completeness directly!

[_hylandSL - not here @_hylandSL]

Perks of Cambridge (UK):
- Original Cambridge
- Near but not actually in London
- Almost entirely flat terrain ideal for cycling
- Option to peer at beautiful university buildings through their railings
- Driest part of England (still rains >100 days/year)

[_hylandSL - not here @_hylandSL]

We are hiring a senior researcher in ML for healthcare at MSR Cambridge (UK)! The position is in my team, so if you get it you will work with me (is this a pro or a con? do not answer). Focus is multimodal (~vision-language) models for radiology! Link: https://t.co/iq39xsVOSF

[CRUK Cambridge Centre @CRUKCamCentre]

Our next Integrated Cancer Medicine and @CambridgeMedAI Seminar a week today!

Multimodal AI for Radiology
Stephanie Hyland, Principal Researcher, @MSFTResearch Cambridge

🗓️ Thur 18 April, 15:00-17:00
📌 Rayleigh seminar room @Maxwell_Centre

Sign up 👉 https://t.co/8G9vwj1doy

[chilconference @CHILconference]

🚨 Deadline Extension Alert 🚨
We’re extending our submission deadline to Feb 16th EoD! 🎉 Don’t miss the opportunity to polish your work & submit the best version of your paper! https://t.co/HKNEAwflBm #CHIL2024 #CallForPapers #ml4h #machinelearning #ai #health #MoreTimeToShine

[MRC Biostatistics Unit @MRC_BSU]

Full programme for our 20th Armitage Workshop is now confirmed. Fantastic line-up of speakers including from @_hylandSL, @ltronneberg, Brendan Murphy, Jessica Barrett & keynote by @amyhherring.

Find out more & register FREE:
https://t.co/2zMtQtWKNs

[_hylandSL - not here @_hylandSL]

(that said, not every internship produces a paper. That's okay - not everything needs to be published, and sometimes/often we do things bigger than 3 month project!)

[_hylandSL - not here @_hylandSL]

Our #TACL paper on Compositional Zero-Shot Domain Transfer is now out! https://t.co/1Lm6yMSL2x
We show that training on general-domain task data (say, NLI) and in-domain unstructured data (say, radiology reports) enables in-domain task capability (radiology NLI)!

[_hylandSL - not here @_hylandSL]

Last note: this paper is an outcome of @hardy_qr's internship last summer. If you're interested in a research internship with us on machine learning in health/medicine, keep an eye on the MSR careers page (https://t.co/m4rBwM7RAt) or wait for me to tweet more about it.