.png){kind=icon}
.png){kind=icon}
Return
A knowledge learning-driven feature selection framework for survival prediction
.png){kind=icon}
PRE
AI
DOI：10.1016/j.neucom.2026.134554.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Gastric cancer (GC) is one of the leading causes of cancer-related mortality worldwide, and accurate prediction of five-year survival is essential for personalized treatment and clinical decision-making. However, conventional prognostic methods often struggle to capture the complex nonlinear relationships and redundant information embedded in clinical variables. To address this issue, this study proposes a collaborative knowledge learning-based feature selection framework for five-year survival prediction in gastric cancer. Specifically, a collaborative knowledge slime mould algorithm (CKSMA) with adaptive gaining–sharing learning mechanisms is developed to enhance population learning and global search capability. A binary variant, bCKSMA, is further integrated with a support vector machine (SVM) to construct a wrapper-based feature learning model, termed bCKSMA-SVM. Experiments conducted on a clinical dataset of 523 gastric cancer patients demonstrate that the proposed method outperforms conventional machine learning models and existing feature selection approaches. In particular, bCKSMA-SVM achieves an accuracy of 87.9% and an AUC of 0.907 in predicting five-year survival for moderately and poorly differentiated GC patients. Moreover, several clinically important prognostic factors, including vascular invasion, tumor size, lymph node metastasis count, T-stage, hemoglobin level, and leukocyte count, are identified. The results indicate that the proposed collaborative knowledge learning framework can effectively improve feature representation and survival prediction performance, providing reliable support for precision prognosis assessment in gastric cancer.
Journal
IF:
6.5
Papers:
2.5W
Citations:
6.5W
View more organizations