Evolution of Alzheimer's disease research from a health-tech perspective: Insights from text mining

Abstract

Tens of thousands of publications have addressed the global threat posed by Alzheimer's disease (AD), traversing the boundaries of discipline to provide novel insight into the pathogenesis of the disorder and therapeutic options. In the era of big data, text mining can facilitate the review of this overwhelming quantity of research. To extract information about recent knowledge of the physical mechanisms and chemical agents implicated in AD progression and treatment, 17,286 abstracts published over the past five years were analyzed using frequency analysis, supervised similarity assessments, and latent Dirichlet allocation (LDA) topic modeling. Abstracts were classified by scope of their respective journals, then compared across fields. Mechanistic keywords pertained to four broad categories. Chemical keywords were also classified, with commercial drugs exhibiting the highest frequency. Nine coherent topics were identified by the fine-tuned LDA model. Implications for cross-disciplinary collaboration in AD research and innovative text mining techniques are discussed.

Keywords

• Alzheimer's disease
• Data science
• Text mining
• Topic modeling
• Information systems
• Health technology