The global race to predict liver cancer using artificial intelligence (AI) is intensifying. In the United States, an AI-based blood test to forecast hepatocellular carcinoma (HCC) was developed last year. This year, Chinese researchers reported in an international journal an AI tool capable of predicting liver cancer recurrence with 82.2% accuracy.

In South Korea, MEDICAL IP is emerging as a standout player. Its AI-driven liver cancer risk prediction solution, DeepFore, integrates CT scan-based biomarkers from patients with chronic hepatitis B and blood data from electronic medical records (EMRs) to estimate the likelihood of developing HCC within the next eight years.

Currently, high-risk patients with chronic hepatitis B are advised to undergo ultrasound and blood tests every six months, which primarily provide a snapshot of their current condition. DeepFore, however, predicts disease onset in advance, enabling personalized risk management and preventive care.

Notably, the AI stratifies patients into four risk categories—minimal, low, intermediate, and high. This allows clinicians to prioritize necessary tests for high-risk patients while reducing unnecessary procedures for low-risk individuals, thereby improving the efficient allocation of medical resources.

In May, DeepFore received Class 3 medical device approval from the Korean Ministry of Food and Drug Safety (MFDS), formally validating its safety and effectiveness. The solution was also highlighted as a reference case in the Digital Medical Device Software Approval and Review Guidelines, showcasing its clinical performance.

Further cementing its credibility, research leveraging this technology was published in the Journal of Hepatology (IF 33.3), serving as the cover article for that issue. It was also cited in the latest European Association for the Study of the Liver (EASL) guidelines for managing chronic hepatitis B.

The guidelines noted that "advanced imaging techniques such as radiomics and artificial intelligence-driven models are expected to refine HCC prediction, optimise surveillance strategies, and integrate into electronic health record systems for automated, risk-based screening protocols."

This acknowledgment underscores DeepFore’s role in overcoming limitations of conventional diagnostic methods and highlights AI’s potential as a guiding force in the future of medicine.

A spokesperson for MEDICAL IP stated, "Beyond our CT-based body composition analysis solution, DeepCatch, we are pioneering a new paradigm in liver cancer prevention with DeepFore, which fuses CT imaging and blood data for precise risk prediction. The post-hepatectomy recurrence prediction tool, DeepFore Recur, has also received MFDS Class 3 approval, demonstrating AI’s potential to support the entire liver cancer care continuum—from risk prediction to prognosis management—setting a new standard in clinical practice."