Discretizing multiple continuous predictors with U-shaped relationships with lnOR: introducing the recursive gradient scanning method in clinical and epidemiological research
The simulation and empirical study demonstrated that the RGS method outperformed other common discretization methods in terms of goodness of fit and predictive ability. However, in the future, we will focus on addressing challenges related to separation or missing binary responses, and we will require more data to validate our method.
Authors
- Fei Yi (ORCID: https://orcid.org/0000-0002-4377-6285)
- Peng Xu (ORCID: https://orcid.org/0000-0002-0999-8546)
- Nanxiang Zhang (ORCID: https://orcid.org/0000-0002-7920-588X)
- Yuduan Han
- Ying Liu (ORCID: https://orcid.org/0000-0002-8997-2257)
- Su H (ORCID: https://orcid.org/0000-0002-1154-8077)
- Jinxin Zhang (ORCID: https://orcid.org/0000-0002-1123-5556)
- Shuo Yang (ORCID: https://orcid.org/0000-0002-6834-4905)
- Yingfeng Ge (ORCID: https://orcid.org/0009-0007-9105-669X)
- Abdullahi Hilowle
Institutions
- Sun Yat-sen University (CN)
- Third Affiliated Hospital of Sun Yat-sen University (CN)
- First Affiliated Hospital of Bengbu Medical College (CN)
Publication Details
- Journal
- BMC Medical Research Methodology
- Published
- 2025-03-12
- DOI
- https://doi.org/10.1186/s12874-025-02522-4
- Primary Topic
- Statistical Methods and Inference
- Type
- article
- Field-Weighted Citation Impact
- 0.00