Abstract
Survival analysis represents an important outcome measure in clinical research and clinical trials; further, survival ranking may offer additional advantages in clinical trials. In this study, we developed GuanRank, a non-parametric ranking-based technique to transform patients' survival data into a linear space of hazard ranks. The transformation enables the utilization of machine learning base-learners including Gaussian process regression, Lasso, and random forest on survival data. The method was submitted to the DREAM Amyotrophic Lateral Sclerosis (ALS) Stratification Challenge. Ranked first place, the model gave more accurate ranking predictions on the PRO-ACT ALS dataset in comparison to Cox proportional hazard model. By utilizing right-censored data in its training process, the method demonstrated its state-of-the-art predictive power in ALS survival ranking. Its feature selection identified multiple important factors, some of which conflicts with previous studies.
Highlights
Survival analysis is essential in clinical research
Common regression algorithms are fitted to the hazard ranks using selected features, and the best prediction model is determined through cross-validation tests
The proposed workflow includes a complete hazard ranking algorithm to address this problem: the ranking algorithm assigns hazard ranks to all subjects, and a machine learning model can be fitted to the hazard ranks
Summary
Survival analysis is essential in clinical research. To assess efficacy and safety of a novel therapy on human subjects, researchers need to carry out clinical trials and collect longitudinal data from participants [1]. To reliably and robustly estimate the effect of a new therapy scheme on subjects’ survival, survival analysis is required. From massive time-event data from clinical trials, survival analysis identifies factors that are predictive to subjects’ lifetime and estimates survival time for new subjects. The most widely used survival prediction model nowadays is Cox proportional hazards (PH) model [3]. Validating the assumptions is important and inappropriate adoption of the Cox model might lead to erroneous predictions [4,5,6]. The original Cox model and its variants have been adopted in all areas to this day, and researchers continue to improve them for different scenarios
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have