Predictive Modeling: What, Why, How

By Ivo Abraham, PhD, RN

Ivo-Abraham-100x100
Ivo Abraham, PhD, RN
Professor of Pharmacy & Medicine
University of Arizona
Tucson, AZ

Ivo Abraham, PhD, a nurse by profession and an outcomes and effectiveness researcher by trade, is a professor of Pharmacy and Medicine at the University of Arizona in Tucson, where he is also affiliated with the Center for Health Outcomes and PharmacoEconomic Research, the Arizona Cancer Center, and the Center for Applied Genomics and Genetic Medicine. Dr. Abraham has served as a regular or visiting professor at universities in the United States, Europe, and Asia. He currently serves as the associate editor for quantitative methods for JAMA Dermatology, and has co-authored more than 350 articles, 75 chapters, and 30 books and monographs.

Dr. Abraham’s research program has been funded continuously since 1984 by governmental agencies, foundations, and corporations worldwide. In the U.S., he has served as an appointed and ad hoc reviewer for the National Institutes of Health, the National Institute for Mental Health, the Agency for Healthcare Research and Quality, and the Veterans Administration. He has also been an expert advisor to the Innovative Medicines Initiative, a joint undertaking of the European Union and the biopharmaceutical industry, to stimulate innovation in human therapeutics.

A native of Belgium, he received his BS (psychiatric nursing) from the Catholic University of Leuven and his MS (psychiatric-mental health nursing) and PhD (clinical research) from the University of Michigan.
  November 20, 2020

Envision a day that cancer clinicians can ask an app to advise on immuno-oncology (IO) treatment options for a patient. That day may not be far off. Big data, deep analytics, and predictive modeling methods are transforming how cancer clinicians weigh treatment options.

The next frontier in IO treatment lies in harnessing highly granular data to explore treatment options by answering questions that will increase the ability of clinicians to deliver personalized medicine. Those questions include:

  1. Which options might apply to this patient, but also which do not, and why?
  2. Why may some options work for this patient, but some not?
  3. What is the relative effectiveness of each of these options for this patient, and how do they stack up against each other?
  4. What adverse events can we expect, when, and why—for this patient?
  5. Which treatment options can this patient tolerate physically and psychologically, and why?
  6. How do treatments compare in balancing effectiveness and safety?
This is where predictive modeling comes in: answering questions about unique patients and their IO treatments so that these treatments can be as individualized as possible.

Read Full Article