Scientists have developed new software algorithms capable of accurately predicting the risk of developing liver cancer based solely on standard medical data.
Cancer Discovery magazine noted that the new algorithms rely on machine learning, analyzing the patient's demographic characteristics, data from electronic medical records, and routine blood test results.
These algorithms were trained using medical data from more than 500,000 people, obtained by scientists from the UK Biomedical Data Bank and from medical data of patients from the United States. The results showed that about 69% of liver cancer cases occurred in people who had not previously had liver disease, highlighting the limitations of current screening methods.
The new software model demonstrated high accuracy in data analysis, outperforming traditional clinical analysis methods. Adding complex genetic and metabolic data to this model did not significantly improve the outcome; adding approximately 15 standard indicators available in routine medical practice was sufficient to provide effective disease prediction.
Researchers believe this software could help doctors identify at-risk patients at an early stage of the disease and direct them to undergo further tests. This is crucial because liver cancer is often diagnosed at a late stage, and early detection greatly increases the chances of successful treatment.
