Sam Abbott
@seabbs.bsky.social
Real-time infectious disease modelling. Developing methods for outbreak response, surveillance, and pandemic preparedness. samabbott.co.uk
Come join me on the epinowcast forum: https://community.epinowcast.org/latest
Come join me on the epinowcast forum: https://community.epinowcast.org/latest
I had some fun vibing up a tool to look at grant amount vs time spent applying for them today. For this one I think it looks like 20-40% or so of the grant pool is spent (due to people working before the grant) on writing the grant vs doing useful work.
gist.github.com/seabbs/d0f61...
gist.github.com/seabbs/d0f61...
October 24, 2025 at 5:47 PM
I had some fun vibing up a tool to look at grant amount vs time spent applying for them today. For this one I think it looks like 20-40% or so of the grant pool is spent (due to people working before the grant) on writing the grant vs doing useful work.
gist.github.com/seabbs/d0f61...
gist.github.com/seabbs/d0f61...
Along with collaborators both in academia and public health we validated this approach, evaluated a range of permutations and compared it to nowcasting approaches used by UKHSA.
August 16, 2025 at 5:55 PM
Along with collaborators both in academia and public health we validated this approach, evaluated a range of permutations and compared it to nowcasting approaches used by UKHSA.
To address these two gaps, @kaitejohnson9.bsky.social led the development of an R package, called baselinenowcast which implements an empirically-driven chain ladder method
August 16, 2025 at 5:55 PM
To address these two gaps, @kaitejohnson9.bsky.social led the development of an R package, called baselinenowcast which implements an empirically-driven chain ladder method
Delays in reporting of epidemiological events can create a misleading impression of declining trends in recent data because recent counts of these events are under-represented.
Correcting this is known as nowcasting
Correcting this is known as nowcasting
August 16, 2025 at 5:55 PM
Delays in reporting of epidemiological events can create a misleading impression of declining trends in recent data because recent counts of these events are under-represented.
Correcting this is known as nowcasting
Correcting this is known as nowcasting
Screenshot of cases from the U.S. CDC’s website showing the number of national measles cases indexed by rash onset date.
Are they going down?
Are they going down?
August 16, 2025 at 5:55 PM
Screenshot of cases from the U.S. CDC’s website showing the number of national measles cases indexed by rash onset date.
Are they going down?
Are they going down?
New CensoredDistributions.jl release adds support for using Turing.jl conditioning on weighted distributions. Unlocks full data simulation from the inference model.
August 13, 2025 at 2:36 PM
New CensoredDistributions.jl release adds support for using Turing.jl conditioning on weighted distributions. Unlocks full data simulation from the inference model.
New CensoredDistributions release adding support for exponentially tilted distributions along with a short tutorial into what that means and why it matters.
August 5, 2025 at 9:50 PM
New CensoredDistributions release adding support for exponentially tilted distributions along with a short tutorial into what that means and why it matters.
Want to try it? Demo with plots:
using Pkg
Pkg.activate(temp=true)
Pkg.add(["CensoredDistributions", "Distributions", "UnicodePlots"])
include(download("https://gist.github.com/seabbs/3cbd4e5bcdba30deec3081ba3838d556/raw/censoreddistributions-demo-gist.jl"))
using Pkg
Pkg.activate(temp=true)
Pkg.add(["CensoredDistributions", "Distributions", "UnicodePlots"])
include(download("https://gist.github.com/seabbs/3cbd4e5bcdba30deec3081ba3838d556/raw/censoreddistributions-demo-gist.jl"))
August 3, 2025 at 10:48 PM
Want to try it? Demo with plots:
using Pkg
Pkg.activate(temp=true)
Pkg.add(["CensoredDistributions", "Distributions", "UnicodePlots"])
include(download("https://gist.github.com/seabbs/3cbd4e5bcdba30deec3081ba3838d556/raw/censoreddistributions-demo-gist.jl"))
using Pkg
Pkg.activate(temp=true)
Pkg.add(["CensoredDistributions", "Distributions", "UnicodePlots"])
include(download("https://gist.github.com/seabbs/3cbd4e5bcdba30deec3081ba3838d556/raw/censoreddistributions-demo-gist.jl"))
🧵 Just released CensoredDistributions.jl! Born from the 2024 Epistorm Rt collaborathon in Boston, this package tackles messy outbreak data where you know something happened within a time window, but not exactly when.
August 3, 2025 at 10:48 PM
🧵 Just released CensoredDistributions.jl! Born from the 2024 Epistorm Rt collaborathon in Boston, this package tackles messy outbreak data where you know something happened within a time window, but not exactly when.
This is what I got running /read-up in a local package (primarycensored). Pretty spot on.
July 17, 2025 at 7:38 PM
This is what I got running /read-up in a local package (primarycensored). Pretty spot on.
Looking forward to meeting you if you are on the course. Also time to get hyped this is my raffle prize (the steer was to bring uni merch or a text book you had written)
July 15, 2025 at 10:44 AM
Looking forward to meeting you if you are on the course. Also time to get hyped this is my raffle prize (the steer was to bring uni merch or a text book you had written)
I was so inspired by the meeting I had to come for an after work drink at the harbour side.
June 18, 2025 at 5:28 PM
I was so inspired by the meeting I had to come for an after work drink at the harbour side.
Turns out it’s a nice place for a workshop. Had a lovely walk/run/swim and didn’t fall off a cliff. Science time now.
April 28, 2025 at 10:36 AM
Turns out it’s a nice place for a workshop. Had a lovely walk/run/swim and didn’t fall off a cliff. Science time now.
This is an example (the recursion fun times are on me and not James).
April 7, 2025 at 2:00 PM
This is an example (the recursion fun times are on me and not James).
In particular I love these caller graphs which would be so nice in pkgdown/roxygen.
Note someone is amazing at naming functions.
Note someone is amazing at naming functions.
March 19, 2025 at 3:35 PM
In particular I love these caller graphs which would be so nice in pkgdown/roxygen.
Note someone is amazing at naming functions.
Note someone is amazing at naming functions.
To round things out here is a taster of some of the figures in the docs
March 7, 2025 at 9:04 PM
To round things out here is a taster of some of the figures in the docs
This is really cool stuff. Only skimmed so far but looking forward to reading it properly tomorrow. Love this figure.
February 5, 2025 at 7:25 PM
This is really cool stuff. Only skimmed so far but looking forward to reading it properly tomorrow. Love this figure.
December 6, 2024 at 11:59 AM
Another good example from our ecosystem is {EpiEstim} which has more than 10 years of largely (entirely?) unfunded support from @anne-cori.bsky.social and others.
github.com/mrc-ide/EpiE...
github.com/mrc-ide/EpiE...
December 6, 2024 at 11:08 AM
Another good example from our ecosystem is {EpiEstim} which has more than 10 years of largely (entirely?) unfunded support from @anne-cori.bsky.social and others.
github.com/mrc-ide/EpiE...
github.com/mrc-ide/EpiE...
Another thing I like to think about is how much work I expect a package to need. For example, our new {primarycensored} package is essentially a small backend tool for the ecosystem so I would expect long term much less code churn despite it still being maintained.
github.com/epinowcast/p...
github.com/epinowcast/p...
December 6, 2024 at 11:08 AM
Another thing I like to think about is how much work I expect a package to need. For example, our new {primarycensored} package is essentially a small backend tool for the ecosystem so I would expect long term much less code churn despite it still being maintained.
github.com/epinowcast/p...
github.com/epinowcast/p...