Forecasting age-specific breast cancer mortality using functional data models.

Area of Science:

• Biostatistics
• Epidemiology
• Public Health

Background:

• Accurate breast cancer mortality forecasts are vital for resource allocation and evaluating screening programs.
• Understanding age-specific mortality trends is crucial for effective breast cancer control strategies.

Purpose of the Study:

• To model age-specific breast cancer mortality time trends using functional data analysis.
• To forecast age-specific mortality functions utilizing a state-space approach.

Main Methods:

• Applied functional data analysis to Australian breast cancer mortality rates (1921-2001) across age groups.
• Utilized non-parametric smoothing, principal components analysis, and a state-space model for forecasting.
• Employed an automatic modeling framework for exponential smoothing method selection.

Main Results:

• Breast cancer mortality rates stabilized from 1960-1990s, followed by a recent decline.
• Four basis functions explained 99.3% of the variation in mortality curves.
• Twenty-year forecasts indicate a continuing but slower decline, stabilizing post-2010, with the greatest reduction in older women.

Conclusions:

• The study provides a robust method for forecasting breast cancer mortality trends.
• Forecasts suggest a positive outlook with declining mortality across all age groups.
• The methodology can be extended to include covariates like hormone replacement therapy and mammographic screening impacts.