Abstract:
Machine learning and artificial intelligence have become a reality in clinical medicine. Very recently, artificial intelligence has exceeded human diagnostic accuracy in cardiology, dermatology, ophthalmology and radiology. Clinical informaticists need to have a working knowledge of these modalities because they are being used for predictive analytics, clinical decision support, image recognition, voice recognition and natural language processing.

The standard pathway to learn machine learning is through a Masters-level data science program, specifically learning one of the programming languages (R or Python). While this approach is optimal, it is not practical for those not in a degree program and there is a very steep learning curve associated with programming languages. In addition, knowledge of higher math (calculus and linear algebra) is generally required.

An alternate approach towards “democratizing machine learning” is through the use of machine learning software. This workshop will discuss seven such programs but focus only on RapidMiner which is felt to be the “best of breed.” This software package automates many of the data preparation, exploration and visualization phases (TurboPrep), as well as the modeling phase (AutoModel).

Workshop participants will learn how to perform data preparation, exploration and analysis using this platform. They will download datasets to predict heart disease (classification) and medical charges (regression). The machine learning software will automatically select multiple appropriate algorithms and then compare algorithm performance with standard measures of accuracy.

Describe the new knowledge and additional skills the participant will gain after attending your presentation.: Attendees will be able to understand the difference between supervised and unsupervised learning. They will use the data science platform RapidMiner to prepare, explore, visualize and analyze two health-related data sets. They will be able to create models for classification and regression.

Authors:

Robert Hoyt (Presenter)
Virginia Commonwealth University