For the first time, scientists have applied machine learning, a form of artificial intelligence (AI), to identify the factors most closely linked to cancer survival in nearly every country across the globe.
The research, published in the leading cancer journal Annals of Oncology, goes beyond broad comparisons to show which specific policy changes or system improvements could have the greatest impact on cancer survival in each nation. The team has also created an online tool that allows users to select a country and see how factors such as national wealth, access to radiotherapy, and global health coverage relate to cancer outcomes.
Turning Global data Into Practical Insights
Dr. Edward Christopher Dee, a resident physician in radiation oncology at Memorial Sloan Kettering (MSK) Cancer Center in New York, USA, and a co-leader of the study, highlighted why the work matters. “Global cancer outcomes vary greatly, largely due to differences in national health systems. We wanted to create an actionable, data-driven framework that helps countries identify their most impactful policy levers to reduce cancer mortality and close equity gaps.”
He noted that several factors consistently stood out. “We found that access to radiotherapy, universal health coverage and economic strength were often vital levers being associated with better national cancer outcomes. However, other key factors were relevant as well.”
Analyzing Cancer and Health System Data From 185 Countries
to reach these conclusions, Dr. Dee and his colleagues used machine learning to examine cancer incidence and death data from the Global Cancer Observatory (GLOBOCAN 2022), covering 185 countries. They combined this facts with health system data gathered from the World Health Organization, the World Bank, United Nations agencies, and the Directory of radiotherapy Centres.
The dataset included health spending as a percentage of GDP, GDP per capita, the number of physicians, nurses, midwives, and surgical workers per 1000 people, levels of universal health coverage, access to pathology services, a human growth index, the number of radiotherapy centers per 1000 people, a gender inequality index, and the share of healthcare costs paid directly by patients.
Building the Machine Learning Model
The machine learning model was developed by Mr. Milit Patel, the study’s first author.He is a researcher in biochemistry, statistics and data science, healthcare reform and innovation at the University of Texas at Austin, USA, and at MSK.
Mr. Patel explained the reasoning behind this approach. “We chose to use machine learning models because they allow us to generate estimates – and related predictions – specific to each country. We are, of course, aware of the limitations of population level data but hope these findings can guide cancer system planning globally.”
Measuring Cancer Care Effectiveness
The model calculates mortality-to-incidence ratios (MIR), which represent the share of cancer cases that result in death and serve as an indicator of how effective cancer care is in a given country. To show how individual factors influence these estimates, the researchers used a method that explains predictions by measuring each variable’s contribution, known as SHAP (Shapley Additive exPlanations).
According to Mr.Patel, the goal was to move from description to action. “Beyond simply describing disparities, our approach provides actionable, data-driven roadmaps for policymakers, showing precisely which
