WHAT ARE THE EVOLUTION PATTERNS OF RESEARCH STRANDS IN WATERBORNE DISEASE TRANSMISSION IN GROUNDWATER?

Groundwater is associated with a significant fraction of all waterborne disease outbreaks. An estimated 750,000 to 5.9 million illnesses per year result from contaminated groundwater in the US. Mortality from these illnesses may be 1400 – 9400 deaths per year. The major contaminants from groundwater which may cause disease include microbial pathogens, heavy metals, and organic chemicals. The study of disease transmission in groundwater has been developed for decades and extraordinary achievements have been published in scientific journals. The objective of this research is to examine the innovative patterns and issues in groundwater disease research by implementing literature mining with both text and visual analytics. The analysis is based on 462 papers (1971 – 2017) from PubMed medical literature database within the scope of groundwater disease. An article text analysis and visualization model developed by Ji et al. (2015) is used in the study. Article similarities are calculated based on five article elements: title, abstract, MeSH (Medical Subject Headings), publication type, and author. Semi-supervised and unsupervised machine learning approaches are then used for article classification and clustering. The resulted patterns are visualized as a 2D article map with the distribution of 11 article clusters. Next, corresponding cluster topics are determined using a natural language processing tool (NLTK) and a keywords extraction tool (RAKE). Yearly publication numbers are analyzed for each cluster. As a result, the golden period of groundwater disease research is after 2000. The study of arsenic disease is most popular across the 11 topics. Arsenic disease, bacteria/virus related disease, and chemical exposures are more diverse comparing to other topics. New topics including risk assessment, QMRA technique, contaminants leaching in the soil, and research in Bangladesh are under rapid development after 2010.