Exploring machine learning techniques using patient interactions in online health forums to classify drug safety

This dissertation explores the use of personal health messages collected from online message forums to predict drug safety using natural language processing and machine learning techniques. Drug safety is defined as any drug with an active safety alert from the US Food and Drug Administration (FDA). It is believed that this is the first exploration of patient derived data of this type for pharmacovigilance – the study of drugs once released to market for safety. It is believed that this is the first application of machine learning and natural language processing techniques to be used for pharmicovigilance on patient derived data. We present results demonstrating the identification of drugs withdrawn from market as well as predictions of other potential safety alert drugs. One example includes Meridia, a weight loss drug linked with death for those with cardiovascular disease. The drug is identified based on data presented two years before FDA and European Union (EU) advisory panels were formed and the subsequent withdrawal of the drug from market within the EU and United States.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Library & Information Science

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2012

Author dc:creator

• Chee, Brant

Contributors dc:contributor

• Schatz, Bruce R.
• Gasser, Les
• Karahalios, Karrie G.
• Blake, Catherine

• Machine learning
• Natural language processing
• Adverse drug events
• Pharmacovigilance

Statement dc:rights

• Copyright 2011 Brant Chee

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/29787

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/29787