Multimodal AI model significantly enhances HCC detection accuracy

A multimodal artificial intelligence model demonstrated impressive results for hepatocellular carcinoma (HCC) detection. Evaluations from 1,092 participants across training, internal validation, and external validation cohorts showed the model’s area under the curve (AUC) values reaching 0.985 and 0.915. Notably, it excelled in identifying small HCCs and those negative for alpha-fetoprotein, achieving AUCs of 0.996 and 0.974, respectively. Thus, integration of clinical, radiological, and immunological data markedly improved detection rates.