To improve risk prediction for gastrointestinal (GI) cancers using machine learning (ML) techniques, particularly in the context of highly imbalanced cohort data.
Key Findings:
GI cancers accounted for nearly 5 million new cases and over 3 million deaths globally in 2022, according to [source].
The study identified 156 incident GI cancer cases (2.0%) among the 7652 participants.
PCUSTe showed promise in preserving population structure in highly imbalanced cohorts.
Interpretation:
Limitations:
The study's findings may not be generalizable beyond the South Korean population.
High costs and resource-intensive nature of prospective cohort studies limit broader application.
