Analyzing Global Cancer Control: Progress of National Cancer Control Programs through Composite Indicators and Regression Modeling

Abstract

Aim: Cancer is a significant public health concern, and National Cancer Control Programs (NCCPs) are crucial for reducing its burden. However, assessing the progress of NCCPs is challenging due to the complexity of cancer control outcomes and the various factors that influence them. Composite indicators can provide a comprehensive and accurate assessment of NCCP progress. Materials and Methods: The dataset was compiled for 144 countries and comprised eight composite indices and two high-level comparative indicators (mortality-to-cancer incidence ratio [MIR] and 5-year cancer prevalence-to-incidence ratio [PCIR]) representing NCCP outcomes. Two large databases and six annual composite index reports were consulted. Linear regression analysis and Pearson correlation coefficients were used to establish a relationship between indicators and NCCP outcomes. A multiple regression machine learning model was generated to further improve the accuracy of NCCP outcome prediction. Results: High-income countries had the highest cancer incidence, whereas low-income countries had the highest MIR. Linear regression analysis indicated a negative trend between all composite indicators and MIR, whereas a positive trend was observed with PCIR. The Human Development Index and the Legatum Prosperity Index had the highest adjusted R2 values for MIR (0.74 and 0.73) and PCIR (0.86 and 0.81), respectively. Multiple linear regression modeling was performed, and the results indicated a low mean squared error score (−0.02) and a high R2 score (0.86), suggesting that the model accurately predicts NCCP outcomes. Conclusions: Overall, composite indicators can be an effective tool for evaluating NCCP, and the results of this study can aid in the development and keeping track of NCCP progress for better cancer control.

Keywords

• cancer
• machine learning
• oncology
• radiotherapy
• regression