JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Integrating Machine Learning and the SHapley Additive exPlanations (SHAP) Framework to Predict Lymph Node Metastasis in Gastric Cancer Patients Based on Inflammation Indices and Peripheral Lymphocyte Subpopulations

  • 0Department of Gastroenterological Surgery, Harbin Medical University Cancer Hospital, Harbin, People's Republic of China.
Journal of Inflammation Research +

|

November 28, 2024

|

November 28, 2024

View abstract on PubMed

Summary

This summary is machine-generated.

Accurately predicting lymph node metastasis in gastric cancer is challenging. An XGBoost model using inflammatory index and lymphocyte subgroups, enhanced by SHAP, shows promise, identifying PLR as a key risk factor.

Area Of Science

  • Oncology
  • Biostatistics
  • Bioinformatics

Background

  • Lymph node metastasis is a critical prognostic factor in gastric cancer.
  • Accurate prediction remains a significant clinical challenge.
  • Preoperative parameters are sought for improved prediction models.

Purpose Of The Study

  • To develop and validate an integrated predictive model for lymph node metastasis in gastric cancer.
  • To leverage machine learning and explainability techniques for enhanced prediction accuracy.
  • To identify key predictive factors from the inflammatory index.

Main Methods

  • Utilized machine learning algorithms, specifically XGBoost, for model development.
  • Employed the SHapley Additive exPlanations (SHAP) framework for model interpretability.
  • Focused on preoperatively obtainable inflammatory index parameters and peripheral lymphocyte subgroups.

Main Results

  • XGBoost demonstrated strong performance, achieving AUCs of 0.705 (training) and 0.695 (test) with a Brier score of 0.218 (test).
  • Platelet-to-lymphocyte ratio (PLR) was identified as the most significant predictor of lymph node metastasis.
  • Six genes (IGFN1, CLEC11A, STC2, TFEC, MUC5AC, ANOS1) showed prognostic value for survival.

Conclusions

  • The XGBoost model effectively predicts lymph node metastasis (LNM) in gastric cancer using inflammation index and lymphocyte data.
  • SHAP analysis provides clear insights into variable contributions to LNM prediction.
  • PLR is a critical risk factor for LNM within the inflammatory index in gastric cancer patients.

Keywords: