Replying to
@olibrendel A misquotation that misrepresents your work warrants a correction. Would be entirely reasonable to email the authors, cc'ing the journal editor. Alternatively, a @PubPeer comment would be in order.
Remote
David C. Norris 🇺🇦 🗽
@dcnorris@scicomm.xyz
mastodon 4.7.3Dose Titration Algorithm Tuning #DTAT #DoseIndividualization #epistemology #philsci #MedicalResearch #oncology #ClinicalTrials #DoseFinding #fedi22 (he/him)
244 Followers
545 Following
17 Posts
Joined November 01, 2022
precautionary R pkg:
Replying to
@lymphomation@mastodon.social
I think Karl you may be referencing the Exceptional Responders Initiative (ERI)?
https://academic.oup.com/jnci/article/113/1/27/5825735
That makes an interesting contrast with my use of the 'exceptional responder' concept in this new work. Really, I'm using the idea in reverse, starting from an approved drug and working backward to show that it is marginal. (This is also, incidentally, a decidedly non-oncology application.)
@lukascbossert@mastodon.social @jameshowell@fediscience.org
Replying to
@lukascbossert@mastodon.social
There is indeed a RESTful API:
https://europepmc.org/RestfulWebService#meths
And your idea to think about this as a CI/CD problem definitely widens my perspective!
@jameshowell@fediscience.org
Reading a paper this morning https://onlinelibrary.wiley.com/doi/10.1002/pds.4295 that follows Nosek & Errington (2017) https://elifesciences.org/articles/23383.pdf in muddling the distinction between #replicability as a broader scientific aim vs #reproducibility as captured in the narrow (purely 'computational') idea of #reproducibleresearch discussed e.g. in a #Guix context here https://guix.gnu.org/cookbook/en/html_node/Reproducible-Research.html.
This mistake is not uncommon, and leads (in this present paper, at least) to insufficient stress on computational reproducibility as a sine qua non of any higher-order quality attribute.
Replying to
@davidho@mastodon.world Would be great if we could do that without also funding scientism and methodolatry.
https://discourse.datamethods.org/t/where-are-the-exceptional-responders/17961/13?u=davidcnorrismd
Replying to
@cmconseils Alt text missed the best part, which is that the mobile phone is possum-sized.
Replying to
@jameshowell@fediscience.org
As Karl Popper said, "We do not know: we can only guess." But at the very least, I feel obliged to 'show my work', and to maintain reproducibility as a basis for continual improvement.
@lukascbossert@mastodon.social
Would anyone have advice on FOSS tooling for reproducible biomedical literature searches? What I'd want to do specifically is to document the *absence* of case reports of exceptional responses to drugs that got FDA accelerated approval. (As for why, see https://osf.io/59sya/files/q6trp.) These would start with automated searches of EuropePMC, then annotation of manual review results. Since a new case report could be published at any time, I would need these to be update-able. I'm guessing an approach built on #orgmode #make and other #root technologies could work well. @lukascbossert@mastodon.social @jameshowell@fediscience.org
#reproducibleresearch
Replying to
@khinsen@scholar.social Thanks for your encouragement! Whereas I'd at first thought of Maslow's hierarchy as just a basic pattern to follow, I now think there may be a kind of 'isomorphism' here:
Physiologic Needs: any part of the analytic pipeline (from data to results) that isn't fully scripted à la #reproducibleresearch is effectively DEAD.
Safety Needs: scripts not under source control, unstructured analysis data sets (spreadsheets≈crime), and indeed all unnecessarily imperative pipeline elements (e.g., shell scripts instead of makefiles) create an UNSTABLE and UNSAFE environment for analytic work. (Where applicable, safeguarding sensitive data also belongs in this tier.)
Love & Belonging: code reviews, documentation [incl. tests, pace Dijkstra], preregistration, conference presentations, preprints, and peer review at all stages of the work; using and crediting community-contributed software.
Esteem: contributing reusable code, bug reports, bug fixes, etc. to a community of scientific software users; contributing thoroughly reproducible and criticizable analyses in peer review.
Self-Actualization: At this stage, the values achieved have to be one's own, I think. Examples might be achieving technical transcendence by doing the whole project in #CommonLisp, #Guix or (for me) ISO #Prolog 🧘, or formally proving correctness of algorithms. Another example would be that others extend & improve upon your work. Finally, #scicomm could be regarded as a true pinnacle.
Replying to
@khinsen@scholar.social A well though-out Maslow's Hierarchy style graphic to convey this idea would be useful to me right now, putting [computational] reproducibility as a basic 'physiologic' need.
https://en.wikipedia.org/wiki/Maslow's_hierarchy_of_needs
I also see this 'feature grid' type of treatment https://co-analysis.github.io/coding-guide/workflow/rap/#code-maturity which however does not underscore the #reproducibleresearch concept quite as I'd like.
Do you use any graphical summaries of this kind in your talks?
Replying to
@noodlemaz Thank you for making this effort, Marianne. I'm dismayed to see such careless responses regarding a matter this important and serious.
Replying to
@rachaelspooky@cyberpunk.lol @mullvadnet@mastodon.online This makes me feel very 'hardcore', thank you! 🐇
Town election today, with open positions ≥ candidates in every race. Voted for everybody except one 'AI' consultant running for School Committee. Wrote in 'Hugh Mannity' instead.
#noai
"Market associations, religious welfare networks, professional guilds, neighbourhood councils – these exist everywhere in Pakistan, self-financed and genuinely representative of those they serve.
Somehow, donors never found them. Or rather, they looked right past them. What donors funded instead, quite generously, were those groups established specifically to attract their money."
https://www.theguardian.com/global-development/2026/may/16/aid-programmes-donors-pakistan-ngos-civil-society
#civilsociety
Replying to
@rachaelspooky@cyberpunk.lol How does this compare to the @mullvadnet@mastodon.online Browser?
Replying to
@zotero@fosstodon.org Not sure if it was there in the old version, but upgrading (which took < 2 minutes) helped me see the top-right corner drop-down V for the open tabs: