Using AI to identify high-risk liver cancer patients

An AI model was better than standard statistical models at predicting whether those with liver cirrhosis were likely to develop liver cancer, according to a new study. Scientists retroactively looked at VA data in nearly 11,000 individuals who had hepatitis C-related cirrhosis and who went on to develop liver cancer by the three-year follow-up, and found that a neural network-based model was better able to discriminate between those who were likely to develop cancer and those who weren't — allowing it to find more cases — and also had higher accuracy scores for predicting these results. The scientists also found that the AI model could be used to predict those with the highest risk of developing cancer so that these patients could be prioritized for treatment.