image: Figure 6. Mapping underexplored connections in the aging research literature through semantic overlap analysis. (A) Heatmap of average TF-IDF score of the top 20 most significant words from each cluster when evaluated against documents in every other cluster using the dataset containing all documents. Rows represent the source clusters from which the top 20 words were selected based on their TF-IDF score. Columns represent the target clusters where the mean TF-IDF scores of these words were computed. Color represents the magnitude of the average TF-IDF score. (B) Top 3 most and least studied relationships among clusters (all documents). (C) Heatmap of average TF-IDF score of the top 20 most significant words from each BoA cluster when evaluated against documents in every other BoA cluster using the dataset containing only BoA-related clusters. Rows represent the source clusters from which the top 20 words were selected based on their TF-IDF score. Columns represent the target clusters where the mean TF-IDF scores of these words were computed. Color represents the magnitude of the average TF-IDF score. (D) Top 3 most and least studied relationships among BoA clusters.
Credit: Copyright: © 2025 Perez-Maletzki and Sanz-Ros. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
“This study outlines shifting priorities and translational gaps in aging research and offers a scalable, data-driven alternative to conventional reviews.”
BUFFALO, NY — December 23, 2025 — A new research paper was published in Volume 17, Issue 11 of Aging-US on November 25, 2025, titled “A natural language processing–driven map of the aging research landscape.”
In this study, Jose Perez-Maletzki from Universidad Europea de Valencia and Universitat de València, together with Jorge Sanz-Ros from Stanford University School of Medicine, used artificial intelligence (AI) to analyze a century of global aging research, revealing shifts in focus and highlighting underexplored areas.
The team analyzed over 460,000 scientific abstracts published between 1925 and 2023 to identify key themes, trends, and research gaps in the study of aging. Their goal was to provide a comprehensive, unbiased view of how the field has evolved and where future research could have the greatest impact.
The study found that aging research has moved from basic cellular studies and animal models to a growing focus on clinical topics, particularly age-related diseases such as Alzheimer’s and dementia. Using natural language processing and machine learning, the researchers grouped publications into thematic clusters and tracked how interest in each topic changed over time.
“By integrating Latent Dirichlet Allocation (LDA), term frequency-inverse document frequency (TF-IDF) analysis, dimensionality reduction and clustering, we delineate a comprehensive thematic landscape of aging research.”
One key finding was the growing separation between basic biological studies and clinical research. While both areas have grown significantly, they often progress independently with limited overlap. Clinical studies tend to focus on geriatrics, healthcare, and neurodegenerative diseases, while basic science emphasizes cellular mechanisms such as oxidative stress, telomere shortening, mitochondrial dysfunction, and senescence. The authors note that this lack of integration limits the translation of laboratory discoveries into medical applications.
The study also showed that some emerging topics, such as autophagy, RNA biology, and nutrient sensing, are expanding rapidly but remain separated from clinical applications. In contrast, long-established links, such as those between cancer and aging, remain strong. The analysis also highlighted that potentially important associations, such as those between mitochondrial dysfunction and senescence or epigenetics and autophagy, are rarely studied and may be new research opportunities.
This AI-driven analysis offers a new way to guide future research by identifying how different areas of aging science are interconnected or isolated. It also highlights how research priorities may be shaped by policy or funding trends, as seen in the heavy focus on Alzheimer’s disease.
As the global population continues to age, understanding how biological processes relate to clinical outcomes is critical. This study not only offers a historical map of aging science but also serves as a tool to support more connected, interdisciplinary, and effective future research.
Paper DOI: https://doi.org/10.18632/aging.206340
Corresponding author: Jorge Sanz-Ros – jsanzros@stanford.edu
Abstract video: https://www.youtube.com/watch?v=O4dJUGQ2ZcU
Keywords: aging, literature mining, natural language processing, topic modelling, synthesis
Click here to sign up for free Altmetric alerts about this article.
______
To learn more about the journal, please visit www.Aging-US.com and connect with us on social media:
- Bluesky
- ResearchGate
- X
- YouTube
- Spotify, and available wherever you listen to podcasts
Click here to subscribe to Aging-US publication updates.
For media inquiries, please contact media@impactjournals.com.
Journal
Aging-US
Method of Research
News article
Subject of Research
Not applicable
Article Title
A natural language processing–driven map of the aging research landscape
Article Publication Date
25-Nov-2025
COI Statement
The authors declare no conflicts of interest. Large language models were employed to assist with coding within the Google Colab environment.