In one session, Patrick Ryan introduced the amazing platform OHDSI. During the discussion, we tried to build a prediction model to predicting HCC in patients with chronic HepB patients in 5 years. We have tried the prediction model using more than 10 000 individuals and found that the AUC is 0.84! This is very exciting because compared to those reported in literature, our study is the best so far in terms of both sample size as well as prediction performance.
So the next step is external validation! We have already found one, but the more the better. If you are interested in this study and have relevant data, please reply or drop me an email (liu.shi@psych.ox.ac.uk).
Thank you and I’m looking forward to hearing from you!
I am Linguo Li from Fudan University. I collaborated a Longitudinal cohort including both HBV infection and no HBV infection population and studies the genetic mechanism of HCC. I would like to join the study and see what could I contribute.
The IQVIA team is happy to externally validate your model on some of our data sets. Can you send us a link to your protocol and study package? We will kick off the process to have the protocol reviewed by our data owners. @sseager will have someone from her team run the package. We should be able to share results in a couple of weeks (may be a little slow with US holiday this week).
Thanks for your email. It will be great to have your external validation. We haven’t generated the detail on protocol yet, but we will let you know ASAP.
I’m an epidemiologist working at IDIAPJGol, a primary care research institute in Barcelona, Spain. I work with the SIDIAP database which includes electronic health records mainly from primary care for nearly 6 million people.
We are still in the process of mapping SIDIAP to the OMOP-CDM, but depending on the information needed to externally validate you model, we could try it.
Please let us know when you can share the study protocol with us so we can check if it is feasible to participate with our data or not.
@shiliuswgch - Just a friendly nudge, no protocol means we can’t even start asking for permission to run the study. When you get a chance, please finalize this so we can at least start the process.