AI Maps Hidden Forces Shaping Cancer Survival Worldwide
- Scientists have used machine learning to pinpoint factors most closely linked to cancer survival rates in nearly every country worldwide.
- The research, published in Annals of Oncology, identifies specific policy changes and system improvements that could improve cancer survival in each nation.
- Edward Christopher Dee, a radiation oncology resident at Memorial Sloan Kettering Cancer Center in New York, explained the study's purpose. "Global cancer outcomes vary greatly due to differences...
Scientists have used machine learning to pinpoint factors most closely linked to cancer survival rates in nearly every country worldwide.
The research, published in Annals of Oncology, identifies specific policy changes and system improvements that could improve cancer survival in each nation. Researchers also created an online tool allowing users to explore how factors like national wealth, radiotherapy access, and global health coverage correlate with cancer outcomes.
Dr. Edward Christopher Dee, a radiation oncology resident at Memorial Sloan Kettering Cancer Center in New York, explained the study’s purpose. “Global cancer outcomes vary greatly due to differences in national health systems.We wanted a data-driven framework to help countries identify the most impactful ways to reduce cancer deaths and address inequities.”
He added that access to radiotherapy, universal health coverage, and economic strength frequently correlated with better outcomes, but other factors also played a role.
To reach these conclusions, Dr. Dee and colleagues used machine learning to analyze cancer incidence and mortality data from the Global Cancer Observatory (GLOBOCAN 2022), covering 185 countries. They combined this with health system data 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 healthcare workers per 1000 people, levels of universal health coverage, access to pathology services, a human advancement index, the number of radiotherapy centers per 1000 people, a gender inequality index, and out-of-pocket healthcare costs.
Milit Patel, a researcher at the University of Texas at Austin and MSK, developed the machine learning model. He explained the approach: “Machine learning models allow us to generate country-specific estimates and predictions. We recognize the limitations of population-level data, but hope these findings can guide global cancer system planning.”
The model calculates mortality-to-incidence ratios (MIR), indicating how effective cancer care is in a country. Researchers used SHAP (Shapley Additive exPlanations) to measure each variable’s contribution to these estimates.
According to Patel,the goal is to move beyond simply identifying disparities. “Our approach provides actionable, data-driven roadmaps for policymakers, showing which health system investments will have the greatest impact for each country. As the global cancer burden grows, these insights can definitely help nations prioritize resources.”