A cutting-edge AI model aids in forecasting the likelihood of acute liver damage worsening.

Japanese researchers recently published their research findings in the new issue of the American academic journal Proceedings of the National Academy of Sciences. They developed an artificial intelligence (AI) model that can predict whether the condition of patients with acute liver damage will deteriorate into acute liver failure based on information such as blood test results at the time of their first visit to the hospital. Japanese institutions including Kyushu University and Nagoya University recently released a joint press release saying that about 99 percent of patients with acute liver damage do not need treatment and can recover naturally, while about 1 percent of patients will develop liver function decline and progress into acute liver failure. The treatment of acute liver failure caused by acute liver damage is difficult, and the number of patients who fall ill each year is not large, so it is difficult for inexperienced medical institutions to take appropriate measures in a timely manner, which directly affects the survival rate of patients. The researchers analyzed the changes in the coagulation factor prothrombin (a blood coagulation factor produced in the liver) in 319 patients with acute liver damage during the first week after hospitalization. They found that acute liver damage patients can be divided into six groups according to the clinical course and prognosis. Group 1 and Group 2 only need to observe the course of the disease and can recover naturally; Group 3 and Group 4 respond to internal medicine treatment and can recover after treatment; Group 5 and Group 6 do not respond to internal medicine treatment and need to undergo liver transplantation or eventually die. The study showed that the application of AI technology can accurately predict which group the patient will enter based on information such as blood test results at the time of the first visit to the hospital. The accuracy rate of predicting whether it will enter Group 5 and Group 6, which requires transplantation, is about 90 percent, and the accuracy rate of predicting whether it will enter Group 3 and Group 4 is about 80 percent.