Adverse event report (AER) data are a key source of signal for post marketing drug surveillance. The standard methodology to analyze AER data applies disproportionality metrics, which estimate the strength of drug/side-effect associations from discrete counts of their occurrence at report level. However, in other domains, improvements in predictive modeling accuracy have been obtained through representation learning, where discrete features are replaced by distributed representations learned from unlabeled data. This paper describes aer2vec, a novel representational approach for AER data in which concept embeddings emerge from neural networks trained to predict drug/side-effect co-occurrence. Trained models are evaluated for their utility in identifying drug/side-effect relationships, with improvements over disproportionality metrics in most cases. In addition, we evaluate the utility of an otherwise-untapped resource in the Food and Drug Administration (FDA) AER system -reporter designations of suspected causality -and find that incorporating this information enhances performance of all models evaluated.
Thought disorder (TD) as reflected in incoherent speech is a cardinal symptom of schizophrenia and related disorders.Quantification of the degree of TD can inform diagnosis, monitoring, and timely intervention. Consequently, there has been an interest in applying methods of distributional semantics to quantify incoherence of spoken language. Prior studies have generally involved few participants and utilized speech data collected in on-site structured interviews. In this paper we conduct a comprehensive evaluation of approaches to quantify incoherence using distributional semantics, including a novel variant that measures the global coherence of text. This evaluation is conducted in the context of “audio diaries” collected from participants experiencing auditory verbal hallucinations using a smartphone application. Results reveal our novel global coherence metric using the centroid (weighted vector average) outperforms established approaches in their agreement with human annotators, supporting their preferential use in the context of short recordings of unstructured and largely spontaneous speech.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.