THURSDAY, April 2, 2026 (NewsDay News) — A machine learning model can predict the risk of hepatocellular carcinoma using commonly available data, according to a study published online March 26. Cancer discovery.
Jan Klusman, MD, from the RWTH Aachen University Hospital in Germany, and colleagues used multimodality data from more than 900,000 people with 983 cases of HCC across the UK Biobank study (development cohort) and the All Us Research Program (to develop an external testing framework for HCC risk studies). The individual and collective contributions of data methods, including demographics, lifestyle, health records, blood, genomics, and metabolomics, were evaluated.
The researchers found that on internal and external test sets, the final random forest-based models performed significantly better than any publicly available risk score. Robustness was demonstrated across ethnic subgroups, allowing for a more comprehensive interpretation.
“Our study highlights the potential of a simple and easy-to-use machine learning model to improve risk stratification for HCC using only routinely collected clinical data,” said co-senior author Carolyn W. Schneider, MD, also of the RWTH Aachen University Hospital, said in a statement. “If validated in additional populations, our model will allow primary care physicians to more effectively identify at-risk patients and refer them for liver cancer screening. This may allow for earlier detection of patients with this aggressive disease and improved outcomes.”
Several authors disclosed ties to the biopharmaceutical industry.
