Related Experiment Video
Updated: Jan 13, 2026
08:50
Predictive Immune Modeling of Solid Tumors
Published on: February 25, 2020
7.4K
Development and validation of a predictive model for extranodal natural killer/T-cell lymphoma
Shuo Li1,2, Li-Min Gao3, Huang-Ming Hong4
1State Key Laboratory of Oncology in South China; Guangdong Provincial Clinical Research Center for Cancer, Sun Yat-sen University Cancer Center, Guangzhou, Guangdong, China.
NPJ Digital Medicine
|January 6, 2026
Summary
A new machine-learning (ML) model, the ENKTL-ML score, improves survival prediction for extra-nodal natural killer/T-cell lymphoma (ENKTL). This algorithm offers greater accuracy than existing models for better patient stratification.
Area of Science:
- Oncology
- Medical Informatics
- Machine Learning
Background:
- Extra-nodal natural killer/T-cell lymphoma (ENKTL) survival prediction models currently lack sufficient accuracy.
- Clinical and pathological factors are crucial for prognostication in ENKTL.
Purpose of the Study:
- To develop and validate a machine-learning (ML) algorithm for improved survival prediction in ENKTL.
- To compare the performance of the developed ML model against established prognostic indices.
Main Methods:
- Analysis of clinical and pathological data from 977 ENKTL patients across four cohorts.
- Development and evaluation of 16 ML algorithms, with Gradient Boosting Machine (GBM) selected for the final model.
- Performance assessment using Harrell's c-index, ROC curves, calibration curves, and decision curve analysis (DCAs).
Main Results:
- The Gradient Boosting Machine (GBM) algorithm demonstrated superior performance, leading to the ENKTL-ML score.
- The ENKTL-ML score achieved high c-indexes in evaluation (0.82) and external validation (0.84, 0.83) cohorts.
- The model effectively stratified patients into three distinct survival outcome groups and outperformed IPI, KPI, PINK-E, and NRI models (P < 0.001).
Conclusions:
- The ENKTL-ML score provides a more accurate and reliable method for predicting survival in ENKTL patients compared to existing models.
- This ML-driven score can aid clinicians in making more informed treatment and management decisions.
- An online calculator is available to facilitate the clinical application of the ENKTL-ML score.
