JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Construction and interpretation of weight-balanced enhanced machine learning models for predicting liver metastasis risk in colorectal cancer patients

  • 0School of Information Management, Wuhan University, Wuhan, Hubei, 430072, People's Republic of China.
Discover Oncology +

|

February 12, 2025

|

February 12, 2025

View abstract on PubMed

Summary

This summary is machine-generated.

Machine learning models can predict colorectal cancer (CRC) liver metastasis. Elevated carcinoembryonic antigen (CEA) and specific treatment strategies are key predictors, guiding personalized patient care.

Area Of Science

  • Oncology
  • Medical Informatics
  • Machine Learning in Healthcare

Background

  • Colorectal cancer (CRC) is a leading cause of cancer mortality.
  • Liver metastases significantly worsen CRC patient prognosis.
  • Predicting liver metastasis is crucial for timely intervention.

Purpose Of The Study

  • Develop and evaluate machine learning (ML) models to predict liver metastasis in CRC patients.
  • Aid clinical decision-making for optimal patient management.
  • Identify key predictors of liver metastasis in CRC.

Main Methods

  • Retrospective analysis of CRC cases from the SEER database (2010-2015).
  • Development and comparison of six ML models: LR, RF, GBM, XGBoost, CatBoost, LightGBM.
  • Model optimization using a weight-balancing algorithm and tenfold cross-validation.
  • SHAP analysis for model interpretability and external validation using a 2018-2021 cohort.

Main Results

  • The CatBoost model demonstrated superior performance (AUC 0.8844, recall 0.8060, F1-score 0.6736).
  • Key predictors identified: elevated carcinoembryonic antigen (CEA), systemic therapy, N and T stages, and chemotherapy.
  • Systemic therapy increased risk in N0 stage but was beneficial with lymph node metastasis.
  • Preoperative radiation therapy was more effective than postoperative.

Conclusions

  • Elevated CEA is a critical predictor of CRC liver metastasis.
  • Systemic therapy in lymph node-negative patients increases metastasis risk.
  • Preoperative radiation therapy is more effective than postoperative for metastasis control.
  • Optimizing treatment strategies based on patient characteristics is essential.

Keywords: