Harnessing machine learning and multivariate analysis to explore global trends in Cannabis sativa research

TL;DR

This study uses data science to analyze global Cannabis sativa research trends from 1974 to 2024, revealing geographic and thematic patterns.

Contribution

A novel framework combining LDA and HJ-Biplot methods for analyzing and visualizing research trends in Cannabis sativa.

Findings

Medical applications of Cannabis sativa are increasingly studied in North America and Europe.

Research gaps persist in Africa and South America.

The combined method offers a replicable approach for managing large bibliographic datasets.

Abstract

This study employs advanced data science techniques to explore global research trends in Cannabis sativa from 1974 to 2024. This research integrated bibliographic datasets from PubMed, Scopus, and Web of Science. By combining latent Dirichlet allocation (LDA) and HJ-Biplot methods, we extracted actionable insights from large-scale data to address the current gap in long-term global research monitoring. The analysis identified key research topics, geographic disparities, and temporal trends, providing a comprehensive overview of the evolution of Cannabis sativa studies. The results highlight an increasing focus on the medical applications of Cannabis sativa, particularly in North America and Europe, while highlighting research gaps in emerging regions such as Africa and South America. Furthermore, the integration of multivariate methods with machine learning offers a replicable framework for managing large bibliographic datasets and enhancing data-driven decision-making in research management. Additionally, combining topic modeling with multivariate visualization provides a novel framework to understand how research themes evolve and interact. This approach serves as a strategic tool for stakeholders navigating the rapidly changing cannabis field.

Figures