Sunday, October 23, 2011: 2:00 PM
Soldier Field (Hyatt Regency Chicago)
Course Type: Half Day
Course Level: Beginner

Format Requirements: This is both a conceptual and a hands-on experience course. No prior knowledge of artificial intelligence is required. Participants will require an Intel based laptop running Windows XP or higher. Illustrative ANN models models will be constructed using MATLAB's Neural Network toolbox. Although some familarity with MATLAB is a plus, it is not a must. A guest license for MATLAB and the Neural Network toolbox will be provided to course participants.

Background: Artificial Neural Networks (ANNs) are computer models capable of processing a large amount of information simultaneously by learning from previous cases. Inspired by the structure and function of biological neural networks, ANNs contain layers of computing nodes, called neurons, that operate as nonlinear summing devices. ANNs have the power to duplicate aspects of human intelligence while incorporating the processing power of computers. Because of their analytical power, ANNs have been used in many critical medical decision making problems. The applications of ANNs in medical decision making include predicting an output value (e.g., predicting cancer risk), classifying an object (e.g., classifying masses as benign or malignant), approximating a function (e.g., approximating tumor growth rate), and recognizing a pattern in multifactorial data (e.g., computer-aided detection in clinical problems). This course will introduce interested investigators to basics of ANNs, their application areas, and implementation through a hands-on experience.

Description and Objectives: This course will provide an introduction to the principles of ANNs and their implementation in medical decision making. The course will start with an introduction to ANNs, their similarities with the biological neural networks and ANNs' working principles. Then, various clinical applications of ANNs will be demonstrated. These examples will be followed by a hands-on experience using MATLAB's Neural Networks toolbox. Finally, the limitations of ANNs and their comparison to other artificial intelligence and statistical methods will be discussed. The course objectives include:

  • Review of Artificial Neural Network (ANN) theory
  • Demonstration of various clinical applications of ANNs
  • Illustration of how to construct and analyze simple ANN models
  • Understanding the advantages and disadvantages of ANNs over other artificial intelligence and statistical models
Course Director:
Turgay Ayer, PhD