danlewer.bsky.social
Consultant in Public Health at Bradford Institute for Health Research. Interested in quantitative research methods and mental health
120 posts
233 followers
177 following
Regular Contributor
Active Commenter
comment in response to
post
Can we please stop searching for biomarkers of aging? We don't need them. It's quite easy to identify people at higher risk of early death.
What we need is to invest in simple, boring, cost effective ways to help prevent illness. I.e. public health and primary care.
comment in response to
post
As a tea drinker, I don't have to put up with all this rubbish
comment in response to
post
I ruminate for weeks and consider quitting research
comment in response to
post
Research question: "If we apply some algorithms to the data, how can we report p<0.05"?
comment in response to
post
Research question: "What is the data telling us?"
comment in response to
post
Wow - you mean we don't have to listen to them burbling on about confounding and bias any more?
comment in response to
post
Lots of people are interested in the use of AI for building prediction models to support early intervention. When I say this is screening and we need to consider potential harms, effectiveness and availability of treatment, etc., they think these concerns are somehow outmoded
comment in response to
post
Thanks Jack 👍
comment in response to
post
What are you thoughts about "minimisation" in the context of clustered RCTs? I'm involved in a cluster RCT that plans to use minimisation - I've read a couple of articles about it but haven't quite got the bottom of the implications for randomisation.
comment in response to
post
Chief Statistics Officer? VP of Statistics (EMEA)?
comment in response to
post
I'm delighted to say our highly secure and productive OpenSAFELY platform has received a further £17 million in funding from Wellcome, including for a new mental health data project. More here, and clips from Today on Radio4 this morning too!
www.bennett.ox.ac.uk/blog/2025/02...
comment in response to
post
100% - it's a bit like headlines that suicide is a leading killer of young people
comment in response to
post
Omg
comment in response to
post
I am worried this might affect time in the opposite direction. We might emerge from the eighth level of significance, believing we had been working on our multidimensional bootstraps for an afternoon - only to find that centuries had passed
comment in response to
post
Could you take it to the third dimension of significance?
comment in response to
post
That's consistent (robust?) evidence of an association approaching significance
comment in response to
post
Isn't the cause and effect in other direction? i.e. you are rich and powerful because you suck up to the local despot?
comment in response to
post
Please feel free to continue!
comment in response to
post
Related - you are required to engage in 'coproduction' of research questions with people who have never done any research, so the coproduced research questions are terrible
comment in response to
post
Tapping the sign
comment in response to
post
If you had an intervention rolled out across 20 different populations at different times, the degree to which those assumptions are wrong would probably just dilute your effect (still bad but I think you're less likely to find random spurious effects)
comment in response to
post
Exactly, and the step change depends on whether you allow a slope change (and vice versa), if you assume the long-term trend is linear, etc. You can't just look at the data points and see, you have to make assumptions and the results depend strongly on those assumptions
comment in response to
post
Loading R
comment in response to
post
Being asked to formulate a research question
comment in response to
post
What does an unusual ECG without other symptoms typically represent and what would you do about it?
comment in response to
post
Let's ask the government if "AI" will help us install RTools please?
comment in response to
post
HOOK ME UP!!
comment in response to
post
Welcome! Even if you are wrong, there are so many benefits to being a late adopter
comment in response to
post
They literally have no idea ... they don't even know what "anonymised" means
comment in response to
post
Actually, it is interesting that they don't mention health, "catching diseases earlier", and "new drugs to cure cancer". That's usually one of the first things they talk about
comment in response to
post
Thank god it won't put more money in the pockets on non functioning people
comment in response to
post
Actual quote: "Backing AI to the hilt can also lead to more money in the pockets of working people"
comment in response to
post
But imagine the rush
comment in response to
post
I enjoyed reading that "Artificial intelligence will be mainlined into the veins of the nation"
comment in response to
post
It's also applicable if you're not treating exposures as time varying and only measure them at baseline - you just have baseline exposures based on information available at cohort entry
comment in response to
post
Brilliant work Tom! I have been saying that the golden rule of retrospective cohort studies is that you must pretend they are prospective, ie. all exposures must be defined based on what you know on that day. That safeguards you against ITB as well as some other problems and misclassifications