Their Words Gave Them Away :A machine learning approach to automatically profiling chronically ill patients based on their natural language use

Keywords
No Thumbnail Available
Authors
Issue Date
2014-01-13
Language
en
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
In this thesis we propose and test a novel approach to automatically assign one of four psychological profiles to chronically ill patients. Using information from a patient's speech and spoken language a machine learning model is built to differentiate between each profile and the other profiles. The predictions of these expert classifiers are gated with a second machine learning model to come to a final decision about the patient's profile. We conclude that for the current problem, a mixture-of-experts consisting of two multi-layered perceptrons and two C4.5 decision trees with a support vector machine as gating mechanism yields the best prediction results. Keywords: machine learning, profiling, speech analysis, language style analysis, computational linguistics
Description
Citation
Faculty
Faculteit der Sociale Wetenschappen