S2-E20 – 14-Variable Machine Learning Model Identifies Probable NASH Patients From Electronic Health Records

Jörn Schattenberg discusses NASHmap, the first Machine Learning model that can identify patients likely to have NASH in clinical settings. Louise, Roger and guest Dr. Kris Kowdley join Jörn to consider the model's benefits from academic, patient treatment and statistical perspectives.

Prof. Schattenberg and colleagues built the NASHmap model from a NIDDK database and validated it using the Optum de-identified EHR dataset. The model includes 14 variables, some obvious, others less so, that produce an AUC of 0.82 in the NIDDK database and 0.79 in the Optum database. The episode focuses on how the model was built and what it implies for clinical trial recruitment, patient care and the view it provides of NASH risk factors. This has powerful implications for future clinical trial recruitment and large-scale patient screening.

TOPICS: Diabetes, Diagnostic Tests, Elastography, FibroScan, NAFLD, NASH, NIDDK, Optum, XGBoost, Machine Learning, Artificial Intelligence, EHR, Electronic Health Records

Surfing The NASH Tsunami is brought to you by HEP Dynamics
#HEPdynamics #SurfingNASH #podcast

Request A Transcript of This Episode
Request Transcript
First
Last

Join The Discussion…

Ask Our Panelists A Question…

Podcast Quesitonaire
Is your question directed towards one of our panelists?

If your question is answered in our next Podcast, would you like be tagged?

Join The Discussion!

Subscribe to our free emails. Receive access to our Discussion Group. Ask your questions directly to our "Surfers"