A multi-country comparison of stochastic models of breast cancer mortality with P-splines smoothing approach

Abstract

Background Precise predictions of incidence and mortality rates due to breast cancer (BC) are required for planning of public health programs as well as for clinical services. A number of approaches has been established for prediction of mortality using stochastic models. The performance of these models intensely depends on different patterns shown by mortality data in different countries. Methods The BC mortality data is retrieved from the Global burden of disease (GBD) study 2017 database. This study include BC mortality rates from 1990 to 2017, with ages 20 to 80+ years old women, for different Asian countries. Our study extend the current literature on Asian BC mortality data, on both the number of considered stochastic mortality models and their rigorous evaluation using multivariate Diebold-Marino test and by range of graphical analysis for multiple countries. Results Study findings reveal that stochastic smoothed mortality models based on functional data analysis generally outperform on quadratic structure of BC mortality rates than the other lee-carter models, both in term of goodness of fit and on forecast accuracy. Besides, smoothed lee carter (SLC) model outperform the functional demographic model (FDM) in case of symmetric structure of BC mortality rates, and provides almost comparable results to FDM in within and outside data forecast accuracy for heterogeneous set of BC mortality rates. Conclusion Considering the SLC model in comparison to the other can be obliging to forecast BC mortality and life expectancy at birth, since it provides even better results in some cases. In the current situation, we can assume that there is no single model, which can truly outperform all the others on every population. Therefore, we also suggest generating BC mortality forecasts using multiple models rather than relying upon any single model.

Epidemiological and sociodemographic transitions of female breast cancer incidence, death, case fatality and DALYs in 21 world regions and globally, from 1990 to 2017: An Age-Period-Cohort Analysis

“…Breast cancer (BC) is the most commonly diagnosed cancer among women and linked with considerable years of life lost (14.9 million DALYs), leading to increased cancer-related morbidity and mortality worldwide [1] , [2] . The burden of the BC is still increasing in form of incidence and mortality both in developing and developed countries characterizing it as one of the most severe burdensome cancer worldwide [3] , [4] . This increasing trend may be an outcome of the obesity epidemic together with dwindling parity [5] .…”

Breast Cancer Mortality Trends and Predictions to 2030 and Its Attributable Risk Factors in East and South Asian Countries

“…The LC-based modeling frameworks are viewed in the current literature as among the most efficient and transparent methods of modeling and projecting mortality improvements. Previously, scholars have used this approach to forecast all-cause and causespecific mortality in different countries (22)(23)(24)(25)(26). The general expression of the model is expressed as: m x,t = e a x +b x +k t +ε x,t , where, m x,t is the central mortality rate at age x and year t, and a x is the average (over time) log-mortality at age x, and b x measures the response at age x to changes in the overall level of mortality over time, and k t represents the overall level of mortality in year t, and ε x,t is the residual term.…”

Handling Overlapping Asymmetric Data Sets—A Twice Penalized P-Spline Approach

“…This research covers a broad range of applications as well as adaptations to the P-Spline method itself, including modifications to the penalty term, as well as the addition of secondary penalty terms to which this work also contributes. P-Splines have been applied within many different domains, including in a medical context such as with Mubarik et al (2020) [14] applying P-Splines to breast cancer mortality data while managing to outperform existing non-smoothing models, and also within a geospatial environment such as with Rodriguez (2018) [15] using P-Splines to model random spatial variation within plant breeding experiments while taking advantage of properties such as a stable and fast estimate and being able to handle missing data and being able to model a non-normal response. P-Splines have also been adapted to be used within a Bayesian context by Lang and Brezger (2004) [16] and are shown in several works to improve predictive modelling, such as with Brezger and Steiner's (2012) [17] work of modelling demand for brands of orange juice and also with Bremhorst and Lambert's (2016) [18] work using survival analysis data.…”