Complete hazard ranking to analyze right-censored data: An ALS survival study

Affiliates

Zhengnan Huang¹, Hongjiu Zhang¹, Jonathan Boss², Stephen A. Goutman³, Bhramar Mukherjee², Ivo D. Dinov^4,5,6, Yuanfang Guan^1,7,8, for the Pooled Resource Open-Access ALS Clinical Trials Consortium

1. Department of Computational Medicine and Bioinformatics, University of Michigan
2. Department of Biostatistics, University of Michigan
3. Department of Neurology, University of Michigan
4. Department of Health Behavior and Biological Sciences, University of Michigan
5. Statistics Online Computational Resource, University of Michigan
6. Michigan Institute for Data Science, University of Michigan
7. Department of Internal Medicine, University of Michigan
8. Department of Electronic Engineering and Computer Science, University of Michigan

Journal

PLoS Computational Biology

Summary

This paper introduces a new method to better analyze survival data from people with ALS, especially when some patients’ outcomes are incomplete or “censored.” The authors created a technique called “GuanRank” that converts survival times into a ranked score, making it easier to apply common machine learning tools that normally don’t work well with this type of data. Using a large ALS dataset, their approach predicted patient survival more accurately than a standard method (the Cox model) and helped identify important factors linked to outcomes, including some that differ from past findings. Overall, the study shows that transforming survival data into rankings can improve prediction and offer new insights into disease progression.

Link to paper

Find the paper here!