JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

A machine learning clinic scoring system for hepatocellular carcinoma based on the Surveillance, Epidemiology, and End Results database

  • 0Department of General Surgery, The Fifth Affiliated Hospital of Zunyi Medical University, Zhuhai, China.
Journal of Gastrointestinal Oncology +

|

July 11, 2024

|

July 11, 2024

View abstract on PubMed

Summary

This summary is machine-generated.

This study developed a machine learning clinical scoring system to predict survival in hepatocellular carcinoma (HCC) patients. The new system accurately stratifies risk, outperforming current staging methods for better HCC management.

Area Of Science

  • Hepatology and oncology research.
  • Development of predictive models in clinical practice.

Background

  • Hepatocellular carcinoma (HCC) is a major global health concern.
  • Limited availability of robust tools for predicting HCC patient prognosis.
  • Need for improved clinical scoring systems for overall survival (OS) and cancer-specific survival (CSS).

Purpose Of The Study

  • To establish a novel clinical scoring system for evaluating OS and CSS in HCC patients.
  • To develop a predictive model using machine learning for HCC prognosis.
  • To compare the efficacy of the new system against the American Joint Committee on Cancer (AJCC) staging.

Main Methods

  • Utilized data from 45,827 primary HCC patients from the SEER Program.
  • Randomly divided patients into training (22,914) and validation (22,913) cohorts.
  • Employed Cox regression, Kaplan-Meier analysis, and machine learning for model development and validation.
  • Assessed model performance using C-index, ROC curves, calibration plots, and clinical decision curve analysis (DCA).

Main Results

  • Identified 11 independent prognostic indicators for CSS and OS.
  • Developed two models demonstrating significant prognostic relevance.
  • The developed models surpassed AJCC staging in predictive accuracy (C-index).
  • Achieved Area Under the Curve (AUC) > 0.74 in time-dependent ROC analyses.
  • Demonstrated superior clinical utility and net benefit compared to AJCC staging via DCA.

Conclusions

  • A machine learning-based clinical scoring system shows strong predictive and risk stratification performance for HCC.
  • The system is readily integrable into clinical practice.
  • Enhances accuracy and efficiency in managing hepatocellular carcinoma patients.

Keywords: