Secondary Logo

Journal Logo

Doc ApprovED

Machine Learning's Potential for Predicting ED Events

Mohseni, Alex MD

doi: 10.1097/01.EEM.0000574828.47496.a5
Doc ApprovED

Figure

Figure

Machine learning is a form of artificial intelligence that allows computers to learn and make predictions without explicit programming. A common form, supervised learning, is a training set of labeled data that is fed to the machine learning model, and then the model identifies patterns and applies unlabeled data to make predictions.

Building a computer vision application might require feeding the model 1000 pictures of animals, for example, each labeled with its species. The model then identifies patterns and labels new pictures with species names with variable rates of success depending on the quality and size of the training data as well as the design of the model itself.

Machine learning and other forms of artificial intelligence are sparking advances in health care, from computer-aided diagnosis of radiology images to predictive modelling of 30-day readmissions. A lot of hype surrounds the term AI, but not all of it is hype. AlphaGo has already beaten the best human Go player, a feat considered impossible just seven years ago because there are more possible Go moves than atoms in the universe.

Machine learning has myriad potential applications in emergency medicine: Predictive modeling of patient arrivals, readmission likelihood, risk of complications, and nonpayment of bills are all obvious applications. Learn more by starting with the Medium essay, “Machine Learning for Humans” by Vishal Maini. (Aug. 19, 2017; http://bit.ly/2W2pcLm.)

Once you are ready to build your first predictive algorithm, try https://rapidminer.com. This site built a well-annotated platform that allows non-coders to input their training data quickly and easily, experiment with different machine learning algorithms, review the results, and then deploy a model. Their tutorials are extensive and fun to use. Best of all, their free version is good for up to 10,000 rows of data. I'm excited to see what you all build! Let me know at emn@lww.com.

Share this article on Twitter and Facebook.

Access the links in EMN by reading this on our website, www.EM-News.com.

Comments? Write to us at emn@lww.com.

Dr. Mohseniis an emergency physician, a telemedicine provider with CirrusMD, and the editor of his own blog, http://CreativeHealthLabs.com. Follow him on Twitter @amohseni, and read his past EMN columns athttp://bit.ly/EMN-DocAPProvED.

Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.