Novel Algorithm Predicts Mortality Risk in Acute-on-Chronic Liver Failure
PATIENTS with acute-on-chronic liver failure (ACLF) typically only have 3 months to live and desperately require a liver transplant for survival. The condition results in severe liver deterioration and multiple organ failure. Unfortunately, liver transplants are hard to come by and receiving one at the right time can be extremely challenging. To add to this complexity, another severely ill patient might take priority. Due to the obstacles in place for patients with ACLF, it would be valuable to be able to precisely predict the mortality risk of patients with ACLF so that they could be prioritised accordingly.
Although there are prediction tools already, current tools do not take into consideration differences in individual symptoms and only apply to patients with similar characteristics such as alcohol use instead. However, it is important to note that a large proportion of patients (40%) have no specific trigger.
Researchers from Southern Medical University (SMU), Foshan, Guangdong, China, aimed to develop a new algorithm that could predict individual mortality. The scientists used data for almost 300 patients from three different hospitals in China. Using their new machine-learning technique called random survival forest (RSF), they were able to successfully predict the mortality risk of individuals over a few months. The team found that they could provide a 95% confident interval using RSF and this tool could be invaluable for predicting individual mortality risk of patients with ACLF.
Following this, the researchers compared their new algorithm to current algorithms, namely the Cox model, and the end-stage live disease model. Interestingly, the team discovered that the RSF algorithm was better than other models for predicting mortality risk and that this tool could help physicians with making a treatment decision for patients with ACLF. The next steps could be to increase the sample size; however, for now, this tool provides much promise and potential and the team have kindly made RSF available online for free.
Zhi-Qio Zhang, SMU, concluded: “This is a rare online web tool that can be valuable in improving treatment decisions for patients due to its ability to predict mortality risk for an individual patient.”