École Nationale Supérieure de
Techniques Avancées

Résumé

This report presents a quantitative and historical analysis of Pharmacology, based on an extensive dataset of over 300,000 scientific works published in the top 50 journals over the past century, using OpenAlex data and the Science of Science framework. The study applied network analysis, BERTopic topic modeling, and machine learning to quantify the field’s evolution and identify factors influencing the impact of academic work. The findings reveal significant structural changes in Pharmacology, including increased collaboration, globalization, and a rising number of references per work, consistent with broader scientific trends. Pharmacology is characterized by a particularly high rate of knowledge obsolescence, indicating substantial shifts in research topics over time. Citation distribution shows a dual pattern (exponential head, power-law tail), supporting the cumulative advantage theory (associated with the “Matthew Effect”). In terms of impact, collaboration alone is not a strong predictor; however, international collaborations and institutional or national prestige are correlated with higher diffusion. Topic modeling uncovered diverse citation dynamics across subfields. While machine learning models demonstrated some predictive power for long-term impact—highlighting author prominence, institutional prestige, topic patterns, reference count, and work type (e.g., review or meta-analysis) as key features—overall predictive performance remained modest.