Using Constrained Information Entropy to Detect Rare Adverse Drug Reactions from Medical Forums

Publisher:
IEEE
Publication Type:
Conference Proceeding
Citation:
38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 2016, pp. 2460 - 2463
Issue Date:
2016
Full metadata record
Files in This Item:
Filename Description Size
07591228.pdfPublished version945.92 kB
Adobe PDF
Adverse drug reactions (ADRs) detection is critical to avoid malpractices yet challenging due to its uncertainty in pre-marketing review and the underreporting in post-marketing surveillance. To conquer this predicament, social media based ADRs detection methods have been proposed recently. However, existing researches are mostly co-occurrence based methods and face several issues, in particularly, leaving out the rare ADRs and unable to distinguish irrelevant ADRs. In this work, we introduce a constrained information entropy (CIE) method to solve these problems. CIE first recognizes the drug-related adverse reactions using a predefined keyword dictionary and then captures high- and low-frequency (rare) ADRs by information entropy. Extensive experiments on medical forums dataset demonstrate that CIE outperforms the state-of-the-art co-occurrence based methods, especially in rare ADRs detection.
Please use this identifier to cite or link to this item: