I received propofol a few hours ago.

It’s wearing off, but I’m still a little overjoyed.

So let’s take the opportunity to be honest with each other.

Look, for example,…

They didn’t give me propofol because I asked for it, like Michael Jackson.

Apparently it was indicated.

I say apparently, because it is not clear.

The anesthesiologist, who I am not saying is not a great professional, asked me:

– Do you want sedation?

– What do you recommend?

– Okay with and without, depending on

– What would you do if I were not a neurologist?

– I’d give it to you.

– Well, come on

– Okay then.

Good.

Occurrence Based Medicine

Whatever comes to mind at the time.

But you know what happens?

It’s not the anesthesiologist’s fault.

Because you can’t have evidence of everything.

Least of all. evidence of everything, applied to every individual.

As much as we would love to, it’s not always possible to have things that you need to do true precision medicine. Things like:

– the full universe of patients
– the durability of the benefit of the intervention
– subpopulations of poor responders
– sufficient follow-up over time
– indirect comparisons between treatments
– surrogate endpoints that allow you to compare even if you don’t have the key variable.

And there are more limitations…

The things you can’t have.

Unless, unless, my internet friend, you…

… You stop creating the registry of the Cooperative Group by pulling the data by hand.

Or stop pulling the database that someone bought from someone you don’t know very well who he is, or where he is, or if he is handsome or ugly.

Stop analyzing by doing your normal statistics, your Kaplan-Meyer, your regression and those little things that you like so much.

Stop all that, 5 years ago you couldn’t, to be honest, but now you can change it.

Using Big Data and machine learning techniques to generate updated data, with flexible variables that you remove and add like a Lego, jumping from hundreds of them to tens of thousands.

Imagine a CRF that:

– Holds thousands of variables
– Fills itself
– Has an “undo button” to change variables as you learn
– Has a function to analyze associations and correlations, to find predictors
– Has a quality checking mechanism that gives you more reliability than a data manager.

So that’s it.

If you want to generate evidence ahead of all the rest who will realize about this in a couple of years, this is it.