Alan Haynes

Statistician working in clinical trials in Switzerland, likes wildlife, ale, OFAH and more

[Alan Haynes @aghaynes]

presize #rstats package up on CRAN!! 😃 Use for precision based sample size calculations. Intended for #clinicaltrials, but also useful for other disciplines #SCTOstatsmethodology @CTUBern

[Alan Haynes @aghaynes]

@R_Graph_Gallery @JohnHelveston if you need an example in favour of base, i had a case recently of using regex to extract text from complex strings. Using stringr was rediculously slow, but regmatches worked well. rewriting the pipeline in base improved performance a lot...

[Alan Haynes @aghaynes]

@ConstantPieters @MatthewBJane gsub("^0", "", sprintf("%2.2f", 0.201)) should work... also in ggplot - scale_x_continuous(..., labels = \(x) gsub("^0", "", sprintf("%2.2f", x))

[Alan Haynes @aghaynes]

@MatthewBJane Gmisc might be useful here perhaps... https://t.co/mLyvMilwAQ

[Alan Haynes @aghaynes]

@ZahraSiddiqui95 if KM curves etc are the issue, ggsurvfit is really nice... https://t.co/b0AKQIFc4l

[Matteo Quartagno @stats_q]

New paper just out: we tend to always use the Hazard Ratio as population-level summary measure in clinical trials with survival outcomes. Is that always justified? [1/N] @tmorris_mrc @IanWhit25399993 @MaxParmarMRCUCL
https://t.co/X6LvLlNPY9

[Alan Haynes @aghaynes]

@r_rabbit_y … |> filter(survived == yes) |> tbl_svysummary(…)? Not sure if this works in a svy context…

[Alan Haynes @aghaynes]

@SEthanMilne @MatthewBJane I think you’re right… adding something like #| fig-height: xxx to the chunk header will probably do it

[Alan Haynes @aghaynes]

@AmyDWillis nope... because the null is removed from the vector...

[Alan Haynes @aghaynes]

@AmyDWillis sapply(x, is.null)?

[Massimo @Rainmaker1973]

Over the course of 3 days scientists pumped 10 tons of cement into an abandoned ant hill. After weeks of digging, the colony’s intricate & impressive structure is revealed. This one

[full video, from "Ants! Natures Secret Power": https://t.co/ol5WJZU6A8]
https://t.co/G71KrCYafP

[Alan Haynes @aghaynes]

@DrAndreaHoward @LucyStats this would be my take on it

[Alan Haynes @aghaynes]

@CedScherer they are now!

[Alan Haynes @aghaynes]

@DaniNembhard @Matt_RogersJa bind_row your two datasets together (harmonise names) and facet_wrap over the model perhaps?

[Alan Haynes @aghaynes]

@tanya_shapiro looks great by the way 🙂

[Alan Haynes @aghaynes]

@tanya_shapiro loading was immediate just now...

[Alan Haynes @aghaynes]

@Jake_Elder52 @IsabellaGhement Mutate recently gained a by argument, so df %>% mutate(gcent.iv = scale(iv), .by = g) should also work. Don't need to worry about ungrouping then either...

[Alan Haynes @aghaynes]

@JWoodelius @giulianonetto No it's just another value that can be calculated from a survival model... https://t.co/Ml78vrUx25

[Alan Haynes @aghaynes]

@giulianonetto @JWoodelius you could use restricted mean survival time (how much time did they survive more within a specific timeframe on average than the other group)

[Alan Haynes @aghaynes]

@statistishdan Found it! ggsurvfit_build is the key! 😄

[Alan Haynes @aghaynes]

@statistishdan Great package! Thanks! Can you use them with patchwork? I've been told the risktable gets removed...?