Risk prediction model for overall survival in lung cancer based on inflammatory and nutritional markers
Abstract This study aims to develop a multidimensional risk prediction model, identify characteristic inflammation-nutrition biomarkers, and optimize clinical decision-making. The study included 500 lung cancer patients diagnosed between October 2019 and October 2024 at a tertiary medical institutio...
Saved in:
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-08-01
|
| Series: | Scientific Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1038/s41598-025-16443-1 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract This study aims to develop a multidimensional risk prediction model, identify characteristic inflammation-nutrition biomarkers, and optimize clinical decision-making. The study included 500 lung cancer patients diagnosed between October 2019 and October 2024 at a tertiary medical institution in Guiyang, China. The exposure variables included eight inflammation-nutrition biomarkers: neutrophil-to-lymphocyte ratio (NLR), lymphocyte-to-monocyte ratio (LMR), platelet-to-lymphocyte ratio (PLR), systemic immune-inflammation index (SII), hemoglobin-albumin-lymphocyte-platelet score (HALP), prognostic nutritional index (PNI), hemoglobin-to-red cell distribution width ratio (HRR), and albumin-to-globulin ratio (ALB/GLB). The outcome variable was overall survival (OS). This study aimed to predict 1-year mortality rather than conduct traditional time-to-event survival analysis. All patients were followed until death or a uniform administrative censoring point.LASSO logistic regression was employed to model the outcome as a binary classification (death within 1 year: yes/no).This study employed a small-sample modeling approach, initially using LASSO regression for feature selection and dimensionality reduction, followed by variance inflation factor and collinearity screening for secondary feature selection. Finally, the Support Vector Machine-Recursive Feature Elimination (SVM-RFE) algorithm was used to optimize feature variables. The results showed that age, clinical stage, poor differentiation, ECOG PS 0–1, serum albumin level, LMR, HRR, and ALB/GLB were independent prognostic factors. Based on these factors, a lung cancer mortality risk prediction model was developed, and a corresponding web-based calculator was created, providing a practical tool to support clinical decision-making and personalized treatment strategies. |
|---|---|
| ISSN: | 2045-2322 |