Projections of cancer mortality risks using spatio-temporal P-spline models

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Accurate cancer mortality risk estimates are crucial for effective public health strategies and resource allocation.
• A time lag exists in the availability of official mortality data, necessitating advanced predictive methods.
• Prostate cancer is a significant public health concern, particularly in Spain.

Purpose of the Study:

• To develop and validate a reliable spatio-temporal model for forecasting cancer mortality and incidence.
• To provide accurate near-future predictions of cancer mortality risks.
• To illustrate the model's application using Spanish prostate cancer data.

Main Methods:

• A spatio-temporal P-spline model was employed to capture smooth temporal trends in mortality and incidence.
• The model was utilized for predicting future cancer mortality and incidence counts across different regions.
• Prediction mean squared error and an appropriate estimator were derived for forecast accuracy assessment.

Main Results:

• The spatio-temporal P-spline model demonstrated effectiveness in capturing temporal trends and predicting cancer counts.
• The study successfully forecasted Spanish prostate cancer mortality for the years 2009-2011.
• The derived estimators provided a measure of the prediction accuracy for the forecasted values.

Conclusions:

• Spatio-temporal P-spline modeling offers a robust approach for near-future cancer mortality and incidence forecasting.
• This methodology can significantly aid in proactive public health planning and resource allocation for cancer control.
• The application to Spanish prostate cancer data highlights the model's practical utility in epidemiological research.