Using Latent Dirichlet Allocation Topic Modeling to Uncover Latent Research Topics and Trends in Renal Cell Carcinoma: Bibliometric Review

JMIR Cancer 2026 AI 6 Explanations View Original
Original Paper (PDF)

Unable to display PDF. Download it here or view on PMC.

Plain-English Explanations
Pages 1-2
Mapping 50 Years of Kidney Cancer Research

This study performed a large-scale analysis of the entire published scientific literature on renal cell carcinoma (RCC), commonly known as kidney cancer, spanning from 1974 to 2023. By examining which topics researchers have studied most and how interest has shifted over time, the authors created a roadmap showing where the field stands and where it needs to go.

The analysis included 35,228 documents from 3,070 different scientific journals, produced by researchers in 118 countries. This remarkable breadth demonstrates how truly global kidney cancer research has become, with an annual growth rate of 9.86% showing that the field is rapidly expanding.

The key tool used was Latent Dirichlet Allocation (LDA), an artificial intelligence technique that reads thousands of papers and automatically groups them into themes or topics based on the words they share. This is similar to how a librarian might sort books by subject, except LDA does it across tens of thousands of documents simultaneously and objectively.

TL;DR: Researchers used AI to analyze 50 years of kidney cancer research across 35,000 papers from 118 countries, revealing which topics dominate the field and which are dangerously under-studied.
Pages 3-5
How AI Reads 50 Years of Science

Latent Dirichlet Allocation (LDA) is a type of machine learning called topic modeling. It works by analyzing the statistical patterns of words across thousands of documents and grouping documents that share similar vocabulary into the same topic. A paper about immunotherapy and a paper about immune checkpoint inhibitors might both fall into a treatment topic, even if they use slightly different terms.

The researchers tested different numbers of topics and determined that 30 topics best described the existing literature without being too broad or too narrow. These 30 topics were then organized by the researchers into 8 major research domains covering the main categories of kidney cancer investigation.

Bibliometric analysis was also used to track how research volume and topic popularity have changed over time. This combined approach of AI topic discovery plus historical trend analysis gives a much richer picture than simply counting publications in each category.

TL;DR: AI topic modeling automatically sorted 35,000 papers into 30 topics and 8 major research domains, then historical trend analysis tracked how interest in each topic has evolved over decades.
Pages 5-8
The 8 Domains Shaping Kidney Cancer Science

The 30 discovered topics were grouped into 8 major research domains: Treatment and Therapies, Biomolecular and Genetic Research, Disease Characteristics, Diagnosis and Evaluation, Metastasis and Dissemination, Epidemiology and Risk Factors, Related Conditions, and Pathological Features. Together, these domains capture the full landscape of kidney cancer science.

The single most prevalent topic across all 50 years was Gene Regulation and MicroRNA, accounting for 5.728% of the literature. This reflects the field's deep investment in understanding the genetic and molecular underpinnings of kidney cancer, particularly how small RNA molecules regulate gene activity in tumors.

The United States was the leading contributor by volume with 10,308 publications, followed by other major research nations. The dominance of a few countries raises questions about whether global patient populations are adequately represented in the research being done.

Trend analysis showed significant growth in areas like immunotherapy and targeted therapies in recent years, reflecting the revolution in cancer treatment brought about by immune checkpoint inhibitor drugs. Topics like surgical techniques and tumor staging have remained consistently active throughout the entire 50-year period.

TL;DR: Kidney cancer research organizes into 8 major domains, with gene regulation being the most studied topic. The US leads in research output, and immunotherapy has seen the most recent growth.
Pages 8-10
The Gaps: What Research Is Missing

Perhaps the most important finding of this bibliometric review is what is not being studied. The authors identified several critically under-researched topics that deserve far more scientific attention given their importance to patients and the future of medicine.

Ferroptosis, a newly discovered form of cancer cell death that is triggered by iron accumulation, is a potentially powerful anti-cancer mechanism that has received almost no attention in kidney cancer research. Similarly, the application of artificial intelligence and machine learning to kidney cancer diagnosis and treatment is vastly under-explored despite AI's proven promise in other cancers.

Long-term evaluation of novel treatment strategies is another major gap. Many new therapies for kidney cancer are approved and adopted based on short-term response data, but long-term survival outcomes and quality-of-life data beyond 5 years are rarely the subject of primary research. This means patients and doctors lack crucial information about what happens over the full course of treatment.

TL;DR: Critical research gaps include ferroptosis as an anti-cancer strategy, AI applications in kidney cancer, and long-term outcome studies for newer therapies, all of which need urgent scientific attention.
Pages 10-11
What This Means for the Future of Kidney Cancer Research

For researchers, this study provides a data-driven guide to where funding and effort should be directed. By showing which topics are saturated with existing research and which are dangerously thin, it helps the scientific community allocate resources more strategically and avoid redundant work.

For patients and patient advocates, this analysis is a reminder that the research agenda is shaped by scientific trends and funding priorities that do not always align with what patients need most. Advocacy efforts can be informed by knowing where the gaps lie, pushing funders to direct money toward under-studied but clinically important questions.

The methodology itself, using AI to map a scientific field, represents a new standard for evidence synthesis. As the volume of scientific literature continues to grow exponentially, tools like LDA become essential for understanding what we collectively know and do not know about a disease.

TL;DR: This bibliometric map gives researchers and patient advocates a roadmap for directing future effort and funding toward the kidney cancer topics that matter most but remain under-studied.
Pages 11-12
A Compass for Kidney Cancer Science

Over five decades, kidney cancer research has grown into a vast global enterprise covering genetics, therapy, surgery, epidemiology, and more. This study is the first to use AI-powered topic modeling at this scale to take stock of all of it and produce an objective, data-driven overview.

The most important conclusion is that while much has been accomplished, the field has significant blind spots. Emerging areas like ferroptosis and AI-based diagnostics need to be brought into the mainstream of kidney cancer research if patients are to benefit from the next generation of breakthroughs.

Ultimately, this work serves as a compass for the scientific community, helping researchers, funders, and clinicians navigate the enormous and complex literature of kidney cancer toward the directions most likely to benefit patients in the years ahead.

TL;DR: Fifty years of kidney cancer research mapped by AI reveals major progress but also critical blind spots in ferroptosis, AI diagnostics, and long-term treatment outcomes that must now be addressed.
Citation: Open Access, 2026. Available at: PMC12810951.