Center for Injury Research and Prevention

Making Machine Learning Accessible to Medical Researchers

August 20, 2019

Earlier this summer, the METACHOP (Math, Engineering, Technology at Children’s Hospital of Philadelphia) group hosted its inaugural Machine Learning Workshop with a goal to make this approach accessible to medical researchers. The event garnered interest from over 100 researchers and led to the promise of three more machine learning workshops to be organized at CHOP to meet this newfound need.

Machine learning is a computational approach designed to find patterns in data that will form reliable and accurate predictive models of specified events, such the occurrence of PTSD in a traumatized child.

With machine learning, specified algorithms search through the space of possibilities contained in the data, to arrive at a predictive model. Afterward the reliability and accuracy of this predictive model are tested on raw data the “machine” has not yet encountered. I have used machine learning for years in my research, including for a study to provide algorithms to help predict and manage certain teen driver behaviors that can contribute to crash risk, such as speed management and inattention.

The event opened with keynote speaker, Ryan J. Urbanowicz,PhD, assistant professor of Informatics at the Perelman School of Medicine at the University of Pennsylvania (pictured below), who talked about the fundamentals of machine learning, including common applications, as well as advice on selecting a machine learning program and evaluation strategies. He offered a foundation for students and researchers to more confidently engage in the opportunities presented by machine learning.

CHOP machine learning workshop

Hands-On Learning

The workshop continued with CHOP Data Scientist Jorge Guerra Marin, MS, conducting a hands-on machine learning session. Participants wrote and analyzed code alongside each other and applied what they learned in Dr. Urbanowicz’s introductory presentation. Using a publicly available biomedical informatics dataset, participants got first-hand experience with developing a decision tree algorithm and designing a machine learning pipeline.

Participants then interacted with the machine learning pipeline and discussed its applications to practical situations. By using a toolset, including Colab notebooks, novices in machine learning could quickly practice without having to download anything to their local machines.

This workshop was developed to meet the need for formal training in big data processing and machine learning for our scientists and engineers in a medical setting. It’s a big part of METACHOP’s goal to keep the CHOP community informed of emerging technology and encouraging collaboration with researchers from the CHOP, University of Pennsylvania and other neighboring research institutions.

If you are interested in upcoming METACHOP workshops and events, please contact me at

**Like what you’ve read? Subscribe to Research in Action to have the latest in child injury prevention delivered to your inbox.**