Immune disease-associated biomarker values are commonly more variable in affected compared to unaffected patient populations, which limits a biomarker’s informative range. Here, the authors formalise a computational solution that splits datasets into informative and uninformative subsets to improve biomarker discovery and performance of multivariate predictive models.
- Gunther Glehr
- Paloma Riquelme
- James A. Hutchinson
