Spatial Survival Analysis
- 602 Downloads
The statistical modeling of censored time-to-event data is the realm of survival analysis; when such data can be located in space and there is scientific interest in understanding the spatial variation in survival outcomes, a spatial survival analysis is performed. The concept of censoring is what makes survival data statistically interesting: for each observation in our dataset, we either observe the time of an event (such as death, disease progression, successful treatment, etc.) or the time or interval of time in which the observation was lost to follow-up. The observations that are lost to follow-up are the censored observations. Spatial survival analysis combines techniques from geostatistics and survival analysis to answer scientific questions like: where in space is the rate of events unusually high (or low)?
Survival analysis per se has a long tradition in the medical sciences, the...
The author is very grateful to Professor Robin Henderson for allowing him to access and use an anonymized version of the leukemia data from Henderson et al. (2002). These data were used to produce figures1–3.
- Banerjee S, Carlin BP (2002) Case studies in Bayesian statistics. Spatial semi-parametric proportional hazards models for analyzing infant mortality rates in Minnesota counties, chapter 6 Springer, New York, pp 137–151Google Scholar
- Cox DR, Oakes D (1984) Analysis of survival data. Chapman & Hall/CRC monographs on statistics & applied probability. CRC Press, London/New YorkGoogle Scholar
- Krige D (1951) A statistical approach to some basic mine valuation problems on the witwatersrand. J Chem Metall Mining Soc S Afr 52:119–139Google Scholar
- Nelson W (1969) Hazard plotting for incomplete failure data. J Qual Technol 1:27–52Google Scholar
- Taylor BM (2015) Auxiliary variable Markov chain Monte Carlo for spatial survival and geostatistical models. Available from http://arxiv.org/abs/ISO1.01665http://arxiv.org/abs/ISO1.01665
- Taylor BM, Rowlingson BS (2014, to appear) spatsurv: an R package for Bayesian inference with spatial survival models. J Stat SoftwGoogle Scholar
- Tonda T, Satoh K, Otani K, Sato Y, Maruyama H, Kawakami H, Tashiro S, Hoshi M, Ohtaki M (2012) Investigation on circular asymmetry of geographical distribution in cancer mortality of Hiroshima atomic bomb survivors based on risk maps: analysis of spatial survival data. Radiat Environ Biophys 51(2):133–141CrossRefGoogle Scholar