Survival analysis of medical database of patients with prostate cancer

V. M. Bure ; E. M. Parilina ; A. I. Rubsha ; L. A. Svirkina

Vestnik Sankt-Peterburgskogo universiteta. Prikladnaâ matematika, informatika, processy upravleniâ, no. 2 (2014), pp. 27-35

Résumé

In the paper survival analysis of the patients with prostate cancer is represented, the purpose of this analysis is the prediction of survival and verification of the effectiveness of treatment. For constructing the survival curves Kaplan–Meier and Cutler–Ederer methods are used. The latter method takes into account censored data. The Cutler–Ederer algorithm is realized with the programming language $C\#$. The software allows to construct the survival curves for medical data. In the paper the hypothesis of homogeneity of the survival curves for different groups of patients are tested. On the basis of available data, it is concluded that the rates of survival in the groups are the same and do not depend on the tumor doubling time. The partition with one day interval was taken as a partition of the time interval in the method. The result of the software work is a plot of the survival curve, as well as the file containing the following information for each interval: days on which the event occurred, namely, the changed behavior of the survival curve, the number of patients followed up to the day in which the event occurred, the number of deaths and censored patients in the appropriate time period, the share of the event, and the cumulative proportion surviving fraction of survivors. Bibliogr. 10. Il. 4. Tables 5.

Détail

Keywords: survival analysis, survival curve, censored data, fitting distribution, Cutler–Ederer method.