Lupus & antiphospholipid syndrome
Development of a least absolute shrinkage and selection operator (LASSO)-based machine learning model for relapse prediction in lupus nephritis
Authors' conclusion
Higher level of 24-hour proteinuria, higher SLEDAI-2K score and low complement C3 level were risk factors for LN relapse, while remission within 12 months was a protective factor. The LASSO-Cox-based predictive model exhibited good performance and may serve as a useful tool for the clinical management of LN.
Study context
- Population
- Patients with lupus nephritis (LN) who received renal biopsies from 2005 to 2024 in Xijing Hospital
- Participants
- 831 patients
- Exposure
- Relapse vs no relapse
- Relapsen = 225
- Outcome
- Renal relapse; performance of the relapse risk prediction model
Methodological detailsprediction model development, single-centre, LASSO-Cox variable selection, multivariate Cox regression, bootstrap resampling, landmark analysis at 12 months
- Prediction model development
- Single-centre
- LASSO-Cox variable selection
- Multivariate Cox regression
- Bootstrap resampling
- Landmark analysis at 12 months
Main finding
As reported in the abstract
Relapse prediction model performance · 12 months postbiopsy (model application)
Relapse
225/831 patients
27.08%
Bootstrap-corrected C-index
0.75
95% CI 0.72 to 0.79
Observational data: associations, not effects.
Further findings
Age of onset
P = 0.001
Proteinuria
P = 0.001
| Outcome | Relapse | Non-relapse | P value |
|---|---|---|---|
| SLEDAI | 12.0 | 7.0 | P = 0.001 |
Nephritis histology
- No significant difference between the two groups
IV methylprednisolone pulse therapy or belimumab
- Proportion significantly higher in the non-relapse group
Original publication
Development of a LASSO-based machine learning model for relapse prediction in lupus nephritis. Lupus Sci Med. Published 25 June 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →