NEW YORK, NY, Dec. 11, 2025 (GLOBE NEWSWIRE) — New data from Marktechpost reveals how machine learning is transforming scientific discovery around the world, and which countries and institutions are driving that change.
A first-of-its-kind analysis of more than 5,000 scientific papers published in the Nature family of journals between January 1 and September 30, 2025 reveals how deeply machine learning (ML) is currently impacting global science.
of ML Global Impact Report 2025covers research from more than 125 countries, identifies the most widely used ML tools in academic research, and highlights the countries and institutions shaping the future of ML-powered scientific discovery.
The findings show that ML is most widely applied in scientific and health research, especially for tasks such as prediction, classification, segmentation, and modeling. These include areas such as early detection imaging and precision diagnostics, genome sequence mapping and mutation tracking, advanced robotics and materials engineering, and large-scale Earth observation analysis.
Asif Razzaq, editor and co-founder of Marktechpost:
“This report shows that machine learning is not only reshaping AI, but science itself. Across imaging, genomics, climate research, and robotics, ML has become a core part of how discoveries are made. The real story here is not hype, but impact. ML is now a fundamental means of modern scientific research.”
Matthew J. Hashim, deputy director of the Artificial Intelligence Institute at the University of Arizona, said:
“Scientific research has moved beyond machine learning experimentation to advancing entire fields through machine learning. Data makes clear that the world's science is now built on shared tools, shared methodologies, and shared collaboration. ML has become one of the most unifying forces in modern research.”
ML in Science: Global Perspectives
As ML becomes a standard part of scientific workflows around the world; The United States is set apart by the wide range of ML technologies adopted across sectors.. According to the report, almost 90% of open source ML tools referenced in scientific research in 2025 will come from the USThis includes many of the world's foundational frameworks used across fields such as imaging, genomics, and environmental science.
at the same time, China is by far the leader in terms of publication volume.,in view of 43% of all ML-enabled papers worldwide – more 2,100 studies China's scientific output reflects a high-density, high-throughput research model, with a large number of ML-based studies published by a relatively concentrated group of leading institutions.
In the United States, it ranks second in total publications. 18%indicating a much more decentralized ecosystem. Universities, hospitals, national laboratories, and private research centers all make meaningful contributions. Harvard Medical School Leads in the number of ML-powered research at US institutions.
Open source ML tools have a US monopoly, but Europe provides some of the most frequently referenced scientific ML models.include Scikit-Learn (France), Unet (Germany), and cat boost (Russia). Other major non-US contributions include: gun and RNN architecture From Canada.
Research amount and density
The report highlights important differences in how countries' research ecosystems expand.
- China Generate massive amounts of ML-enabled science from. A more concentrated group of institutionsgenerates the average 72.8 papers per university (Normalization).
- US shows A larger and more diverse contribution baseaveraging 39.6 papers per institution — reflects the widespread interdisciplinary adoption of ML.
- India and Saudi Arabia It appears as Fast-growing institutional investorsthe footprint of ML-powered research is expanding and participation in collaborative scientific networks is increasing.
These trends suggest that while China leads in output, other countries, including India, Saudi Arabia, and institutions in the United States, are moving beyond traditional centers of scientific leadership and forming new hubs of machine learning-driven innovation.
Collaboration: The backbone of science with ML
Across all regions, Collaboration remains in default model For ML-driven scientific research. Most ML-enabled papers include: 2 to 15 affiliated institutionsIt is often a combination of computational laboratories, specialized research institutes, and medical or industrial partners.
Only a small group of global institutions consistently exist across multiple scientific disciplines and form the backbone of modern AI-driven research. Main research using ML Rarely emanates from a single organization;Instead, we rely on international partnerships that combine software engineering, domain expertise, and experimental science.
Collaboration patterns vary by region.
- China indicates a more concentrated model with mean: 2.6 organizations per paper.
- US Demonstrate a broader network. 4.1 Number of organizations per paperreflecting its deeply collaborative academic environment.
- India, Saudi Arabia, USA We frequently collaborate in applied science, materials research, engineering and computer vision, creating new research corridors across continents.
The hype and real-world impact of neural networks.
Despite the excitement surrounding generative AI, data shows that scientific research is still primarily driven by mature machine learning methods. Traditional ML techniques (such as random forests, SVM, and Scikit-learn workflows) account for 47% of all ML use cases.
When combined with established ensemble and clustering techniques such as GBM, XGBoost, LightGBM, and CatBoost, these traditional approaches account for an overwhelming 77% of ML applications in scientific research.
The majority of scientific research using ML focuses on: Practical, domain-driven goals Rather than state-of-the-art ML innovations, it includes prediction, classification, image segmentation, biological pattern recognition, protein modeling, feature extraction, and more.
About Marktech Post
Marktechpost is a global publication covering artificial intelligence, machine learning, and emerging technology research. The platform highlights advances made by academic institutions, research institutions, and practitioners that are shaping the future of applied AI. https://www.marktechpost.com/
