Boston QSP is postponing our event for Wednesday, March 24, 2020 out of an abundance of caution due to the COVID-19 situation. The new event date for the "Modeling and Simulation in Drug Development" event is TBD.
This MSDD event will feature a talk by Dr. Andy StainStein, Director of Pharmacometrics at Novartis. The talk is titled "Assessing the Uncertainty and Predictive Accuracy Accuracy of Mechanistic Models, with Applications in CML and Target Engagement" and will be followed by a mixer and reception where you can enjoy great company and conversation with fellow community members.
RSVP here to reserve your spot and receive updates regarding this event as well as future Boston QSP events.* Please do not hesitate to RSVP on the waiting list if the RSVP is full as some guests "un-RSVP" as the event gets closer.
Andy Stein is a Director of Pharmacometrics at Novartis in Cambridge, Massachusetts, where he's worked for 10 years. He did postdoctoral research at the Institute for Math and its Applications, a PhD in Applied Mathematics at University of Michigan, and a Bachelor’s and Master’s in Mechanical Engineering at MIT.
Andy will discuss the methods for evaluating the predictive accuracy of mechanistic models. For models that make many predictions over a large number of compounds, benchmarking them against simpler heuristics is a useful way to test if the mechanistic model captures important, relevant biology, or whether it is at risk of overfitting noise in the data. For models that make inferences about a new scenario where there is not a database of similar cases to draw from, we propose the Credibility Assessment Framework developed by Kuemmel et al  together with an Uncertainty Pedigree Table  for assessing the key sources of model uncertainty. We apply these approaches to in two cases. 1) A mechanistic model of Chronic Myeloid Leukemia for prospectively predicting whether patients would be able to discontinue Gleevec therapy. 2) A model for target engagement that includes tissue distribution and shed antigen, for predicting receptor occupancy to help guide dose selection for a monoclonal antibody. A key challenge yet to be addressed is how to explore uncertainty in the choice of structural model itself, when the biology of interest is still under active investigation.
Event Sponsored by:
Venue sponsored by:
About Boston QSP
Boston QSP is a 501(c)(3) non-profit organization whose mission is to foster the sharing of QSP knowledge, challenges, solutions, and opportunities to advance the field as an interdisciplinary community in Boston.