An international team of molecular biologists has developed an innovative AI‑assisted technique for diagnosing liver cancer. The method detects tumours at early stages by analysing DNA fragments that enter the bloodstream from cancerous cells, TASS reports.
The approach was tested on volunteers from two regions of the world, according to the Johns Hopkins University press office. Professor Victor Velculescu noted that earlier studies by his team had shown that analysing DNA fragments in the blood can reveal tumours and chronic liver conditions that increase the risk of cancer.
Given the rising global incidence of liver cancer — linked to chronic diseases, poor diet, and excessive alcohol consumption — early diagnosis is becoming critically important. Velculescu stressed that early detection can save most patients, whereas late diagnosis often leads to poor outcomes.
Four years ago, the researchers made a key breakthrough by identifying DNA fragments released by liver tumour cells — fragments that are absent in healthy individuals. Based on this discovery, they developed an AI algorithm designed to analyse genetic fragments in blood samples to detect tumours.
The method was trialled on 377 volunteers from Romania and Guatemala, 244 of whom had been diagnosed with hepatocellular carcinoma. Researchers collected blood samples and tested whether the neural network could correctly identify the presence of a tumour or rule it out in healthy participants.
The AI system demonstrated high accuracy: it correctly identified the absence of tumours in 92 % of patients and detected cancer in 80 % of the Romanian and Guatemalan participants. This performance significantly exceeds that of existing tests, opening up prospects for large‑scale early screening for liver cancer.
About the Author
