JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

A machine learning approach to differentiate stage IV from stage I colorectal cancer

  • 0Institute of Gastroenterology and Hepatology, Soroka University Medical Center, Faculty of Health Sciences, Ben-Gurion University of the Negev, Beer-Sheva, Israel.
Computers in Biology and Medicine +

|

April 12, 2025

|

April 12, 2025

View abstract on PubMed

Summary

This summary is machine-generated.

Machine learning accurately predicts stage IV colorectal cancer (CRC) using clinical symptoms and lab values like carcinoembryonic antigen (CEA). This approach aids early diagnosis before imaging, improving patient outcomes.

Area Of Science

  • Oncology
  • Medical Informatics
  • Biostatistics

Background

  • Colorectal cancer (CRC) staging is critical for prognosis.
  • Early diagnosis of stage IV CRC significantly impacts patient outcomes.
  • Distinguishing between early and advanced CRC stages remains a clinical challenge.

Purpose Of The Study

  • To develop a novel supervised machine learning approach for diagnosing stage IV CRC.
  • To identify key clinical and laboratory predictors differentiating stage I and stage IV CRC.
  • To compare the diagnostic utility of clinical history and laboratory values in CRC staging.

Main Methods

  • A retrospective study involving 433 patients with stage I CRC and 457 with stage IV CRC.
  • Supervised machine learning, specifically a random forest algorithm, was employed.
  • A decision tree model was utilized for visualizing and identifying critical differentiating factors.

Main Results

  • Symptoms and laboratory values were identified as crucial predictors for stage IV CRC.
  • Change in bowel habits, weight loss, constipation, and abdominal pain were significant indicators.
  • Specific thresholds for carcinoembryonic antigen (CEA) levels (e.g., >260) and combinations with hemoglobin, white blood cell, and platelet counts predicted stage IV CRC.

Conclusions

  • Machine learning effectively identified key predictors for stage IV CRC diagnosis.
  • Clinical symptoms and specific laboratory values (CEA, hemoglobin, WBC, platelets) are vital for early detection.
  • This AI-driven approach can potentially expedite stage IV CRC diagnosis in clinical settings, preceding imaging.

Keywords: