Harrison Wilde

Statistician and data scientist, working @nesta_uk @ucl on Bayesian ML and applications in health and policy, prev. @warwickuni @twitter @datascifellows

[Harrison Wilde @HarrisonDWilde]

@f2harrell Thank you, would be much appreciated

[Prof. Christina Pagel - @chrischirp.bsky.social @chrischirp]

Hancock: "there was no way we could allow the NHS to become overwhelmed"

Except, the NHS WAS overwhelmed

Here is what NHS staff said about that time - Pls read whole 🧵
"Heartbreaking"
"Horrific"
"It broke my soul"
"We cried, we came home exhausted. We were overwhelmed"

1/16

[Prof. Christina Pagel - @chrischirp.bsky.social @chrischirp]

See also this article today from a consultant making the same point. It was awful. It doesn't help future pandemic response to pretend that it was better in the NHS than it was.

https://t.co/yabP711f0Q

[Jeremias Knoblauch @LauchLab]

@vdwild @SGhalebikesabi @sejDino and I rigorously connected Bayesian methods to deep ensembles using Wasserstein gradient flows!
Our results are summarised in a recent NeurIPS oral paper (https://t.co/3gGC4Edvgt), a 30-minute talk (https://t.co/p9RaB2z8o5), and below:

[Harrison Wilde @HarrisonDWilde]

@DS_x_HE Any chance of a recording? I had a clash but was keen to see this :(

[Harrison Wilde @HarrisonDWilde]

@jessRmorley …Totally agree with your broader point though, and very skeptical in general of the claims around this type of solution. I am curious now though as to whether there are some details somewhere on how they could come to these numbers?

[Harrison Wilde @HarrisonDWilde]

@jessRmorley ...I am aware however through personal experience that despite taking great care myself with claims in this type of work, media and journalists very often jump to exceptionalism in headlines etc…

[Prof. Christina Pagel - @chrischirp.bsky.social @chrischirp]

🧵our new definitive @bmj_latest peer-reviewed paper on kids hospitalisations in England!
https://t.co/AOkrjU01Rq

More accessible & shorter version in this @ConversationUK article from me & @katebrown220
https://t.co/nNQnxiFhrR

read on for the highlights! 1/15

[Dr. Deepti Gurdasani @dgurdasani1]

Our @bmj_latest paper led by @HarrisonDWilde out TLDR;
-21000 hosp admissions from COVID (caused or contributed) in children- England from July 2020-Feb 2022
-children<5 yrs & ethnic minorities most at risk
-admissions⬆️during omicron
- 38% would've been classified as CV by JCVI

[Prof. Christina Pagel - @chrischirp.bsky.social @chrischirp]

THREAD: OUR NEW PAPER on vax for 5-11 yr kids.

Most kids unvaxxed but have had at least 1 infection

Do they need vax & what are poss benefits inc vs Long Covid?

Our peer-reviewed paper in @Vaccines_MDPI is here:
https://t.co/SChCctZo7X

READ ON FOR SUMMARY 👇👇 1/26

[Harrison Wilde @HarrisonDWilde]

@SGhalebikesabi Looks cool, are you going Sahra? And do you need to be registered for the AI UK conference or is this a separate free thing, can’t tell from the eventbrite haha

[Harrison Wilde @HarrisonDWilde]

@SolomonMg @CSMaP_NYU Back to the dark side huh, congrats Sol, sounds really cool :)

[Harrison Wilde @HarrisonDWilde]

@chrischirp I want to meet them on Monday!!

[Harrison Wilde @HarrisonDWilde]

@honualx @jadecopet @syhw @adiyossLC Is there any plan to try and compete with more mainstream br mp3, i.e. potential for 192 / 320kbps encodec formulation that sounds indistinguishable from lossless? I'd imagine these tasks become almost exponentially more difficult but I am v intrigued 2/2

[Harrison Wilde @HarrisonDWilde]

@honualx @jadecopet @syhw @adiyossLC I can see there being some great use cases in resource-constrained settings for bitrates this low, I find it easy to detect low br mp3 and to me these super low br encodec outputs sound at least comparable to 64kbps mp3, low quality but in a more adaptive and ignorable sense 1/2

[Harrison Wilde @HarrisonDWilde]

@SGhalebikesabi @MSFTResearch Woo congrats Sahra so well deserved :)

[Harrison Wilde @HarrisonDWilde]

@ecsalomon https://t.co/L0Ne1or4zW DSG’s are sort of that