Missing data imputation and synthetic data simulation through modeling graphical probabilistic dependencies between variables (ModGraProDep): An application to breast cancer survival.

BACKGROUND: Two common issues may arise in certain population-based breast cancer (BC) survival studies: I) missing values in a survivals' predictive variable, such as "Stage" at diagnosis, and II) small sample size due to "imbalance class problem" in certain subsets of patients, demanding data modeling/simulation methods. METHODS: We present a procedure, ModGraProDep, based on graphical modeling (GM) of a dataset to overcome these two issues. The performance of the models derived from ModGraProDep is compared with a set of frequently used classification and machine learning algorithms (Missing Data Problem) and with oversampling algorithms (Synthetic Data Simulation). For the Missing Data Problem we assessed two scenarios: missing completely at random (MCAR) and missing not at random (MNAR). Two validated BC datasets provided by the cancer registries of Girona and Tarragona (northeastern Spain) were used. RESULTS: In both MCAR and MNAR scenarios all models showed poorer prediction performance compared to three GM models: the saturated one (GM.SAT) and two with penalty factors on the partial likelihood (GM.K1 and GM.TEST). However, GM.SAT predictions could lead to non-reliable conclusions in BC survival analysis. Simulation of a "synthetic" dataset derived from GM.SAT could be the worst strategy, but the use of the remaining GMs models could be better than oversampling. CONCLUSION: Our results suggest the use of the GM-procedure presented for one-variable imputation/prediction of missing data and for simulating "synthetic" BC survival datasets. The "synthetic" datasets derived from GMs could be also used in clinical applications of cancer survival data such as predictive risk analysis.

[Excess mortality among breast cancer patients in early stages in Tarragona and Gerona (Spain)]. / Exceso de mortalidad en las pacientes con cáncer de mama en estadios precoces en Tarragona y Gerona (España).

OBJECTIVE: To analyze the population-based survival of breast cancer (CM) diagnosed in early stages estimating the time trends of excess mortality (EM) in the long term in annual and five-year time intervals, and to determine, if possible, a proportion of patients who can be considered cured. METHOD: We included women diagnosed with BC under the age of 60 years in stages I and II in Girona and Tarragona (N = 2453). The observed (OS) and relative survival (RS) were calculated up to 20 years of follow-up. RS was also estimated at annual (RSI) and in five-year intervals (RS5) to graphically assess the EM. The results are presented by age groups (≤49 and 50-59), stage (I/II) and diagnostic period (1985-1994 and 1995-2004). RESULTS: In stage I, OS and RS were higher during 1995-2004 compared to 1985-1994: 3.5% at 15 years of follow-up and 4.5% at 20-years of follow-up. In 1995-2004, the OS surpassed 80% in stage I patients whereas in stage II it remained below 70%. During 1995-2004, the long-term EM did not level off towards 0 (RSI <1) independently of age group, stage and period of diagnosis. After 15 years of follow-up, the 5-year EM oscillated between 1 and 5% in stage I (RS5 ≥0.95) and between 5 and 10% in stage II. CONCLUSIONS: In our cohort, after 15 years of follow-up, it was detected that the annual EM did not disappear and the five-year EM remained between 1 and 10%. Therefore, it was not possible to determine a cure rate of BC during the study period.

[WebSurvCa: web-based estimation of death and survival probabilities in a cohort]. / WebSurvCa: estimación vía web de las probabilidades de fallecimiento y de supervivencia de una cohorte.

Relative survival has been used as a measure of the temporal evolution of the excess risk of death of a cohort of patients diagnosed with cancer, taking into account the mortality of a reference population. Once the excess risk of death has been estimated, three probabilities can be computed at time T: 1) the crude probability of death associated with the cause of initial diagnosis (disease under study), 2) the crude probability of death associated with other causes, and 3) the probability of absolute survival in the cohort at time T. This paper presents the WebSurvCa application (https://shiny.snpstats.net/WebSurvCa/), whereby hospital-based and population-based cancer registries and registries of other diseases can estimate such probabilities in their cohorts by selecting the mortality of the relevant region (reference population).