Abstract
The exponential growth of biomedical and life sciences literature, including research on amyloid biology, has made it increasingly challenging to track new discoveries and gain a comprehensive understanding of the evolution of specific research fields. Advances in natural language models (NLM) and artificial intelligence (AI) approaches now enable large-scale analysis of scientific publications, uncovering hidden patterns and facilitating data-driven insights. Here, a two-dimensional mapping of the global amyloid research landscape is presented, using the transformer-based large language model PubMedBERT, in combination with t-SNE and Latent Dirichlet Allocation (LDA), to analyze more than 140 000 abstracts from the PubMed database. This analysis provides a comprehensive visualization of the amyloid field, capturing key trends such as the historical progression of amyloid research, the emergence of dominant subfields, the distribution of contributing authors and their respective countries, and the identification of latent research topics over time, including chemicals and small molecules. By integrating AI-driven text analysis with large-scale bibliometric data, this study offers a novel perspective on the evolution of amyloid research, facilitating a deeper interdisciplinary understanding. This work serves as a valuable interactive resource for researchers while highlighting the potential of machine learning-driven literature mapping in identifying knowledge gaps and guiding future investigations.