HQ Team
January 14, 2025: A first-of-its-kind machine-learning analysis covering 99 % of the world’s population has identified the exact health-system policies that have helped save the most cancer lives in each of the planet’s 185 nations.
The study, released in the Annals of Oncology and accompanied by an online free public dashboard, turns complex global data into a country-specific “to-do list” for ministries of health, donors and patient advocates.
Researchers fed 20 key variables ranging from GDP per capita and universal-health-coverage (UHC) score to the number of radiotherapy machines per million people into models trained on the latest Global Cancer Observatory (GLOBOCAN 2022) incidence and death figures. The output is a mortality-to-incidence ratio (MIR) for every country and, crucially, a ranked set of “drivers” that explain why some nations lose far fewer patients than others.
Global cancer outcomes factors
“Global cancer outcomes vary greatly, largely due to differences in national health systems,” said co-lead author Dr Edward Christopher Dee, radiation-oncology resident at Memorial Sloan Kettering Cancer Center, New York.
“We wanted an actionable, data-driven framework that helps countries identify their most impactful policy levers to reduce cancer mortality and close equity gaps.”
Three levers dominate the global picture: access to radiotherapy, strength of UHC and economic power.
Yet the model, built by first author Milit Patel, a data-scientist at the University of Texas at Austin and MSK, shows that the fastest route to lower deaths is not the same everywhere.
Brazil, for instance, already scores well on nurse density; the algorithm says expanding UHC would cut the MIR more than adding pathology labs.
Poland, which has boosted insurance coverage in the last decade, should now channel extra funds into radiotherapy machines and GDP growth rather than simply increasing overall health spending, the analysis concludes.
In Japan, the density of radiotherapy centres is the single strongest predictor of better outcomes, while in the United States and the United Kingdom, GDP per capita heads the list, implying that continued economic investment underwrites the high survival rates—81 % of US breast-cancer patients now live at least five years, compared with 66 % in upper-middle-income countries and 46 % in low-income nations, according to CONCORD-3.
China presents a mixed case: rising wealth, UHC and radiotherapy capacity are already paying off, but out-of-pocket costs remain a brake. “High direct costs for patients remain a critical barrier,” the authors warn, even as China’s overall MIR improves.
Interactive map
The interactive map colours each factor green or red to signal where extra effort would yield the biggest immediate drop in deaths. “Green bars represent areas where continued or increased investment is most likely to result in meaningful impact,” Patel explained. “Red bars don’t mean neglect—only that, right now, they are less likely to explain survival gaps.”
Biggest take-away for ministries: target the greens first but keep an eye on reds for equity. With cancer incidence projected to hit 35 million new cases a year by 2050—up from 20 million in 2022—prioritisation is urgent. The World Health Organization estimates that 2.4 million lives could be saved annually if low- and middle-income countries simply achieved the global median MIR.
Strengths of the study include near-universal country coverage, up-to-date inputs and explainable AI (SHAP) that shows exactly how each variable shifts the dial.
Limitations are the usual ecological-data caveats: numbers are national averages, registry quality varies, and the model shows association, not causation.
Still, veteran policy-makers welcomed the tool. “As the global cancer burden grows, this model helps countries maximise impact with limited resources,” Dee concluded. “It turns complex data into understandable, actionable advice—making precision public health possible.”
Users can explore their country’s profile and download policy briefs at the project’s dashboard (Global Health Insights ).
