OHDSI Home | Forums | Wiki | Github

Patient-Level Prediction TC Wednesday May 15th 12:00 EST


(Jreps) #1

Dear All,

In our call on Wednesday we will give a live demo of how you can apply our PLP framework to determine whether an end of observation in US claims is due to death (https://rd.springer.com/article/10.1007/s40264-019-00827-0). Following this we will discuss how we can make our prediction models more user friendly.

Jenna and Peter

(Seng Chan You) #2

Congratulation for the great achievement, @jreps @Rijnbeek .

We’ll validate this model on our database, too.

Then, by using this model, can we solve the problem (underestimated mortality in the US claim data), raised by the recent paper published in JAMA cardiology?

(Chungsoo Kim) #3

Hi, Jenna @jreps
I’m Ted from Ajou univ

I wanted to attend the meeting, but I could not connect.
I don’t know what’s wrong.

Is there a web page where can I watch your live demo video?
I wanna validate your DeadModel in our database.


(Chungsoo Kim) #5

I ran the DEAD model on our DB with chan @SCYou .
For the record, we’ve got AUROC of 92.69 and AUPRC of 99.86, and Brier score of 0.53. Magnificent!

(Selva Muthu Kumaran Sathappan) #6


Is it possible to watch this video online?