Mining Heterogeneous Healthcare Networks Extracted from Health-Consumer-Contributed Contents for Adverse Drug Reaction Detection

Abstract

Since adverse drug reactions (ADRs) represent a serious world-wide health problem, how to detect ADR signals effectively and efficiently has been of great medical significance. Currently, methods proposed to detect ADRs are mainly based on data sources such as spontaneous reporting data, electronic health record, pharmaceutical databases, and biomedical literature. However, these data sources are either limited by high cost, under-reporting ratio, privacy issues, or long publication cycle. In this work, we propose to explore online health communities, a timely, informative and publicly available data source, for ADR detection. We construct a weighted heterogeneous healthcare network that contains drugs, ADRs, diseases, and users extracted from online health consumer-contributed contents, extract topological features, develop weighted path count to quantify the features, and use supervised learning techniques to detect ADR signals. The experiment results show that weighted heterogeneous healthcare network using leverage and lift as weighting schema are more effective in ADR detection than non-weighted counterpart.