Biomedical Text Analytics for Characterizing Climate-Sensitive Disease

Abstract

Large-scale influx of scientific literatures in the biomedical domain, enriched with various biomedical entities like genes, proteins, drugs, diseases, symptoms, microbes, pathogens etc. embed many useful information that remains untapped due to unstructured nature of texts. Processing these texts using NLP techniques, and extracting embedded entities and their relations can provide useful information in creating disease knowledge base, which is a key enabler for the development of effective disease surveillance systems. In this paper, we present a biomedical text analytics approach to identify disease symptoms and relations from biomedical texts for characterizing climate-sensitive diseases. Four climate-sensitive infectious diseases, including Cholera, Dengue, Influenza, and Malaria are considered for experimentation, and it is found that the proposed approach is able to identify new disease symptoms that are even not listed on standard websites like Center for Disease Control (CDC), National Health Survey (NHS), and World Health Organization (WHO). In addition, it also identifies generic relations between diseases and their symptoms. The proposed approach could be useful for the development of climate-sensitive disease surveillance and prevention systems.