zEpid

@zepidpy

Python 3 free-software library with easy to use tools for Follow the code at: by Paul Zivich

Vrijeme pridruživanja: veljača 2018.

Tweetovi

Blokirali ste korisnika/cu @zepidpy

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

  1. Prikvačeni tweet
    28. kol 2018.

    Check out these replications of epidemiology studies using the zEpid python library Includes demonstrations of parametric g-formula and IPW

    Prikaži ovu nit
    Poništi
  2. proslijedio/la je Tweet
    4. velj

    Achieving a full-coverage society means Medicare for All.

    Prikaži ovu nit
    Poništi
  3. 31. sij

    Caveat: g-formula can also be used to solve the above issues. However outcome-model-based approaches don't solve all censoring mechanisms (see Section 8.5 of Hernan & Robins for an example)

    Prikaži ovu nit
    Poništi
  4. 31. sij

    For a discussion of IPCW to correct for informative LTFU, I suggest

    Prikaži ovu nit
    Poništi
  5. 31. sij

    For a published (and more in depth presentation) on the necessity of survey weights for the PACE I would recommend

    Prikaži ovu nit
    Poništi
  6. 31. sij

    While survey weights won't address LTFU, that doesn't justify ignoring survey weights for a non-random sample. Rather we need to use an analytical approach to address informative LTFU & non-random sampling

    Prikaži ovu nit
    Poništi
  7. 31. sij

    Summary: survey weights and censoring weights are meant to solve distinct issues. If you are interested in the population-ACE, and there informative LTFU and the sample is a non-random sample of the target pop. then something needs to be done for each issue

    Prikaži ovu nit
    Poništi
  8. 31. sij

    Recently been involved in a disagreement on the necessity of survey weights. Particularly if there is LTFU then why bother using survey weights. I wrote a simple simulation to demonstrate why survey weights are needed and LTFU is a distinct issue

    Prikaži ovu nit
    Poništi
  9. proslijedio/la je Tweet

    what's wrong with "the problem in American healthcare is the prices" framing 👇

    Poništi
  10. proslijedio/la je Tweet
    27. pro 2019.

    Asked for a comparison of TMLE and double machine learning, this blog post comparing the two approaches in historical context came together quickly

    Poništi
  11. proslijedio/la je Tweet

    Today, the majority of this country, myself included, celebrates the birth of a Brown-skinned Jewish refugee and migrant, a major prophet in Islam, a socialist, a champion of equality, who preached against exorbitant wealth, hunger, homelessness, and hatred. Merry Christmas!

    Poništi
  12. 6. pro 2019.

    Hey what is your go-to estimator of the average treatment effect in your applied work?

    Poništi
  13. proslijedio/la je Tweet
    1. pro 2019.

    Getting a flu shot is praxis, solidarity, mutual aid, etc. — flu is very dangerous for people who can’t get vaccinated!! If you haven’t gotten your shot yet, go do it tomorrow, if not to spare yourself the flu then to protect those of us who can’t get the shot!!

    Poništi
  14. proslijedio/la je Tweet

    In retaliation for 'one-hot coding', I propose we start calling deep learning 'multilevel regression'

    Poništi
  15. proslijedio/la je Tweet

    now that dunking on Gates is a thing can we talk about the fact that he used his billions as leverage to effectively take over health policy for Africa and large parts of Asia and education policy in the US and no one ever voted for him and he’s accountable to literally no one.

    Prikaži ovu nit
    Poništi
  16. proslijedio/la je Tweet

    Say Bill Gates was actually taxed $100 billion. We could end homelessness and provide safe drinking water to everyone in this country. Bill would still be a multibillionaire. Our message: the billionaire class cannot have it all when so many have so little.

    Poništi
  17. 31. lis 2019.

    Happy Halloween!

    Poništi
  18. proslijedio/la je Tweet
    21. lis 2019.

    If I’ve said it once, I’ve said it a 1000 times. Instrumental variables assume no unmeasured confounding!! It’s just the confounding you care about is instrument & outcome, not exposure & outcome. If you don’t believe no unmeasured confounding is ever possible, stop using IVs!!

    Poništi
  19. 3. lis 2019.

    v0.8.1 minor release now available TMLE supports PyGAM's LogisticGAM for custom_models now Note: TMLE will not lose support for ML in v0.9.0 but I am trying to see if I can check whether the input estimators are Donsker and generate a warning if not

    Poništi
  20. 2. lis 2019.

    If you're at come listen to my talk on assortativity by treatment and bias in networks in Ravenna

    Poništi
  21. 19. ruj 2019.

    Why does Word require pixel-perfect accuracy when trying to resize columns/rows? Who thought that was a good idea?

    Poništi

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:

    ·