Journal: Scientific reports
This study evaluates seven machine learning algorithms to improve hepatocellular carcinoma (HCC) detection, focusing on a Filipino cohort, an underrepresented population in HCC research.
Among the models tested, random forest (RF) and LightGBM achieved the highest predictive accuracy (~99%) with strong sensitivity and specificity, using only seven clinical predictors:
- Age
- Albumin
- Alkaline phosphatase
- Alpha-fetoprotein
- Des-gamma-carboxy prothrombin
- Aspartate transaminase
- Platelet count
The findings highlight the robustness of machine learning approaches in HCC detection across diverse risk profiles. They suggest that such models could simplify diagnostics, especially in resource-limited settings, enabling earlier detection and potentially improving patient outcomes.