Clinical Report: Enhancing Risk Assessment for Gastrointestinal Cancers

Overview

This study investigates the use of machine learning techniques to improve risk assessment for gastrointestinal cancers, focusing on addressing class imbalance in cohort data. A novel patient-centered undersampling technique (PCUSTe) is evaluated against traditional resampling methods to enhance predictive performance.

Background

Gastrointestinal cancers represent a significant health burden globally, with millions of new cases and deaths annually. Accurate risk prediction is essential for early detection and targeted prevention strategies, particularly in regions with high incidence rates. Traditional methods face challenges due to class imbalance and the complexity of risk factors, highlighting the need for innovative approaches like machine learning.

Data Highlights

This study utilized data from 12,552 South Korean adults enrolled in the Korea National Cancer Center Screenee Cohort, focusing on diverse predictors for GI cancer risk prediction.

Key Findings

• Machine learning models can improve predictive performance for GI cancer risk by uncovering nonlinear interactions among risk factors.
• Class imbalance in cohort studies poses significant challenges for accurate risk prediction, often leading to poor sensitivity for cancer cases.
• The patient-centered undersampling technique (PCUSTe) was developed to address class imbalance while preserving population structure.
• PCUSTe was compared with established resampling methods like SMOTE and ADASYN, demonstrating its potential advantages.
• Incorporating a wider range of risk factors may enhance the effectiveness of screening and prevention strategies for GI cancers.

Clinical Implications

The findings suggest that machine learning techniques, particularly those addressing class imbalance, could enhance risk assessment for GI cancers. Clinicians may consider integrating these advanced modeling approaches to improve early detection and prevention efforts.

Conclusion

This study highlights the potential of machine learning in refining risk assessment for gastrointestinal cancers, particularly through innovative techniques that address data challenges. Further research is needed to validate these approaches in diverse populations.

Related Resources & Content

1. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries
2. Recommendation: Colorectal Cancer: Screening | United States Preventive Services Taskforce
3. A Cell-free DNA Blood-Based Test for Colorectal Cancer Screening | New England Journal of Medicine
4. Screening and surveillance in patients at high risk for gastric cancer - American Gastroenterological Association
5. Gastric cancer prevention in the Republic of Korea - Population-Based Helicobacter pylori Screen-and-Treat Strategies for Gastric Cancer Prevention: Guidance on Implementation
6. Surgical Endoscopy — Applications of Machine Learning in Surgical Interventions for Upper Gastrointestinal Cancer: A Comprehensive Review
7. Prediction Model Utilizing Machine Learning for Omental Metastasis in Patients with Right-Sided Colon Cancer: A Retrospective Study Across Multiple Centers
8. The ASCO Post — Machine-Learning Model May Predict Unplanned Hospitalizations After Radiation Therapy for Gastrointestinal Cancer
9. The ASCO Post — Machine-Learning Model May Predict Unplanned Hospitalizations After Radiation Therapy for Gastrointestinal Cancer
10. Applications of Machine Learning in Surgical Interventions for Upper Gastrointestinal Cancer: A Comprehensive Review
11. Prediction Model Utilizing Machine Learning for Omental Metastasis in Patients with Right-Sided Colon Cancer: A Retrospective Study Across Multiple Centers
12. Machine-Learning Model May Predict Unplanned Hospitalizations After Radiation Therapy for Gastrointestinal Cancer
13. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries - Gco Iarc
14. Recommendation: Colorectal Cancer: Screening | United States Preventive Services Taskforce
15. A Cell-free DNA Blood-Based Test for Colorectal Cancer Screening | New England Journal of Medicine
16. Screening and surveillance in patients at high risk for gastric cancer - American Gastroenterological Association
17. Gastric cancer prevention in the Republic of Korea - Population-Based Helicobacter pylori Screen-and-Treat Strategies for Gastric Cancer Prevention: Guidance on Implementation - NCBI Bookshelf
18. Diagnosis and Management of Barrett’s Esophagus: An Updated ACG Guideline - PMC
19. LI-RADS US Surveillance Version 2024 for Surveillance of Hepatocellular Carcinoma: An Update to the American College of Radiology US LI-RADS | Radiology
20. ASGE | ASGE Guideline on Screening for Pancreatic Cancer in High-Risk Cohorts
21. Effect of Colonoscopy Screening on Risks of Colorectal Cancer and Related Death | New England Journal of Medicine
22. Effect of invitation to colonoscopy versus faecal immunochemical test screening on colorectal cancer mortality (COLONPREV): a pragmatic, randomised, controlled, non-inferiority trial - ScienceDirect
23. Korean Practice Guidelines for Gastric Cancer 2024: An Evidence-based, Multidisciplinary Approach (Update of 2022 Guideline) - PMC
24. A deep learning algorithm to predict risk of pancreatic cancer from disease trajectories | Nature Medicine
25. Diagnostic Performance of Artificial Intelligence in Detection of Hepatocellular Carcinoma: A Meta-analysis - PMC
26. Diagnostic Risk Prediction Models for Upper Gastrointestinal Cancers: A Systematic Review - PMC
27. GI and Hepatology News - Clinical practice update Ultrasound plus AFP remains the standard for HCC surveillance