Gary Collins  

@GSCollins

Professor of Medical Statistics , Director of UK , , , BMJ Stats Editor, prediction research, cyclist

University of Oxford, UK
ndorms.ox.ac.uk/team/gary-coll…
Vrijeme pridruživanja: svibanj 2011.
56 fotografija ili videozapisa Fotografije i videozapisi

Tweetovi

Blokirali ste korisnika/cu @GSCollins

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @GSCollins

  1. Prikvačeni tweet
    6. kol 2018.

    …and here is a screen shot of two slides from a talk I gave a couple of years ago on the topic - I think says it all (full slides available from here ).

    60 proslijeđenih tweetova 215 korisnika označava da im se sviđa
    Poništi
  2. prije 10 sati

    Interested in improving the (and ) of research reports then visit - comprehensive database of reporting guidelines (for a variety of study designs) and plenty of resources for authors.

    The authors of this paper present a consensus-based checklist to improve and document the transparency of research reports in social and behavioural research.
    6 proslijeđenih tweetova 15 korisnika označava da im se sviđa
    Poništi
  3. proslijedio/la je Tweet
    prije 20 sati

    Another example of research waste - “However, key details were missing about who should act (45%) and when (22%). Specification of who should act was missing in 79% of 15 interventions to reduce duration of treatment in continuing-care wards.” Need Ix descriptions

    Our new paper out today- applying a behavioural approach to specifying interventions to reduce antibiotic prescribing in hospitals
    Prikaži ovu nit
    3 proslijeđena tweeta 9 korisnika označava da im se sviđa
    Poništi
  4. proslijedio/la je Tweet
    4. velj
    Odgovor korisniku/ci

    Calibration of predictions is not overrated. In my opinion, it is underrated as many researchers do not consider it at all. As noted, perfect calibration is hard to achieve across entire range of risks. But we should still examine it.

    1 reply 2 proslijeđena tweeta 5 korisnika označava da im se sviđa
    Poništi
  5. 4. velj

    Steve Goodman "none of these types of reproducibility can be assessed without complete reporting of all relevant aspects of scientific design, conduct, measurements, data, and analysis." . Full and is key for

    12 proslijeđenih tweetova 28 korisnika označava da im se sviđa
    Poništi
  6. proslijedio/la je Tweet
    3. velj

    Excited that this paper is out on what we think it will take to get apps into clinical practice. With and Accompanying editorial by Adam Cohen

    23 proslijeđena tweeta 48 korisnika označava da im se sviđa
    Poništi
  7. proslijedio/la je Tweet
    2. velj

    "digital triage tools are currently not fully clinically validated or tested by product regulators and discovered ‘there is great variation in their clinical performance’"

    1 reply 15 proslijeđenih tweetova 27 korisnika označava da im se sviđa
    Poništi
  8. 2. velj

    Should We Trust Algorithms? by David Spiegelhalter

    4 proslijeđena tweeta 16 korisnika označava da im se sviđa
    Poništi
  9. proslijedio/la je Tweet
    1. velj

    Interim-PET for prognostication in pts w Hodgkin Lymphoma. Great example of a Cochrane Prognostic Factor Review by Aldin et al, using state-of-art methods from protocol, design, searching, data extraction, RoB, analysis, strength of evidence and reporting.

    5 proslijeđenih tweetova 8 korisnika označava da im se sviđa
    Poništi
  10. 31. sij

    For the 46 years and 8 months I've been on this planet, I've been privileged to have been part of the EU (my daughter would not be here without our participation in the Erasmus exchange programme) - in 12 hours time we turn the clocks back to 1973. Sad day!

    1 reply 3 proslijeđena tweeta 40 korisnika označava da im se sviđa
    Poništi
  11. proslijedio/la je Tweet
    30. sij

    High impact pain journals do not promote standards for research transparency. Particularly concerning given evidence of bias in the reporting and synthesis of pain research.

    8 proslijeđenih tweetova 21 korisnik označava da mu se sviđa
    Poništi
  12. proslijedio/la je Tweet
    30. sij

    Am I the only one who’s irked by academic writing advice of “telling a story with your data”? Writing fanfiction about our data is what got us into the reproducibility crisis in the first place. I’d rather have a less compelling narrative than a flowery account of half-truths.

    267 replies 484 proslijeđena tweeta 4.423 korisnika označavaju da im se sviđa
    Poništi
  13. proslijedio/la je Tweet
    30. sij

    thought of the day: Developing predictions from observational data such as informs one about PREVAILING outcomes of PREVAILING treatment strategies and has almost nothing to do with what happens if a DIFFERENT treatment were given to a patient.

    32 proslijeđena tweeta 143 korisnika označavaju da im se sviđa
    Poništi
  14. 30. sij

    Far too many interesting discussions on clinical prediction ( and ) on twitter today. Good sign - that there are lots of people interested on the topic.

    1 proslijeđeni tweet 20 korisnika označava da im se sviđa
    Poništi
  15. proslijedio/la je Tweet
    28. sij

    Interested in search strategy reporting? PRISMA-S coming out soon! Here's the latest version: Full E&E coming soon

    33 proslijeđena tweeta 61 korisnik označava da mu se sviđa
    Poništi
  16. proslijedio/la je Tweet
    30. sij

    4. Net Reclassification Index has been criticized for statistical and interpretation deficiencies; Net Benefit is far more attractive if we want to quantify clinical usefulness

    3 proslijeđena tweeta 34 korisnika označavaju da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  17. proslijedio/la je Tweet
    30. sij

    3. Calibration is the Achilles heel of prediction models; it should be studied at model validation, not meaningful at model development

    6 replies 6 proslijeđenih tweetova 30 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  18. proslijedio/la je Tweet
    30. sij

    2. Showing ROC curves is a waste of space; the AUC is the main measure of interest (paper forthcoming, submitted to )

    6 replies 3 proslijeđena tweeta 29 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  19. proslijedio/la je Tweet
    30. sij

    1. “The authors present AUCs, but to compare models, they need to present the p-value comparing the AUCs”. --> Delta(AUC) at model development should not be tested, the likelihood ratio test is sufficient.

    3 proslijeđena tweeta 31 korisnik označava da mu se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  20. proslijedio/la je Tweet
    30. sij

    Just had a paper rejected that develops and internally validates 2 variants of a prediction model. Rev #1 has 4 key points: 1. Compare AUCs with p-values 2. Show ROC curves 3. Show calibration 4. Show net reclassification index All nonsensical. Let’s go through.

    8 replies 64 proslijeđena tweeta 225 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  21. proslijedio/la je Tweet
    30. sij
    Odgovor korisnicima i sljedećem broju korisnika:

    There a lot of reporting guidelines with different goals :) TRIPOD is for prediction models, though: Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis. At some point, an AI/ML update is coming:

    1 reply 2 proslijeđena tweeta 2 korisnika označavaju da im se sviđa
    Poništi

@GSCollins još nije objavio nijedan Tweet.

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·