Description
Recent improvements in machine learning have led to the ability to generate both complex and sophisticated time-series models for healthcare situations. Though powerful, these models traditionally lack insight into the importance of each of the model’s input values. Here, we introduce a novel element of the Heartwood AnalyticsTM system that ranks both sparse and dense time-series features according to their importance in the model for clear communication of driving factors for complex medical decision making.
Learning Objective: After viewing this submission, the learner should be better able to:
1.) Understand the current challenges and research associated with machine learning model interpretation, especially regarding deep learning models.
2.) Learn the importance of being critical of complicated machine learning technologies in the healthcare space so that they may be used to supplement, rather than replace, decision making.
Authors:
Jacob Ekstrum (Presenter)
CUBRC, Inc.
Noah Poczciwinski, CUBRC, Inc.
Brett Cropp, CUBRC, Inc.
John Coles, CUBRC, Inc.
Presentation Materials: