Advancing Gastrointestinal Cancer Risk Prediction With Patient-Centered Machine Learning: Machine Learning Modeling Study
-
By
-
Daina Baublyte
-
Jeonghee Lee
-
Madhawa Gunathilake
-
Jeongseon Kim
-
June 4, 2026
-
Clinical Scorecard: Enhancing Risk Assessment for Gastrointestinal Cancers Through Patient-Focused Machine Learning: A Study on Modeling Techniques
At a Glance
| Category | Detail |
|---|
| Condition | |
| Key Mechanisms | Machine learning techniques for risk prediction and imbalance mitigation strategies. |
| Target Population | |
| Care Setting | |
Key Highlights
- Nearly 5 million new GI cancer cases and over 3 million deaths reported in 2022.
- Class imbalance in cohort studies poses challenges for accurate risk prediction.
- Patient-centered undersampling technique (PCUSTe) was evaluated to address class imbalance.
- The study included 12,552 participants, with 156 incident GI cancer cases identified.
Guideline-Based Recommendations
Diagnosis
- GI cancers defined according to ICD-10 classifications.
Management
- Utilization of machine learning models to improve risk prediction.
Monitoring & Follow-up
- Follow-up of participants for up to 14 years to track cancer development.
Risks
- High class imbalance in cohort studies can lead to poor sensitivity for cancer cases.
Patient & Prescribing Data
Participants from the Korea National Cancer Center Screenee Cohort.
Focus on identifying high-risk individuals for targeted preventive measures.
Clinical Best Practices
Related Resources & Content
