Outlier detection in survival analysis: an application to ovarian cancer data