Recently, the hepatobiliarypancreatic surgery team of the First Affiliated Hospital of University of South China, as the first corresponding author, published a clinical research achievement titled “Online interpretable dynamic prediction models for clinically significant posthepatectomy liver failure based on machine learning algorithms: A retrospective cohort study” in the form of an article online inInternational Journal of Surgery, an international well-known academic journal (in the Q1 of JCR and Q2 of CAS, IF=15.3) in the field of surgery.

Posthepatectomy liver failure (PHLF) refers to severe liver damage caused by various reasons after hepatectomy. Hepatic functions, such as synthesis, detoxification, excretion,biotransformation, suffer severe impairment, presenting a clinical syndrome primarily manifested by coagulation disorders, jaundice, ascites, hepatic encephalopathy, etc. Itais an inevitable, seriouspotentially life-threatening complication after hepatectomy,also represents a leading cause of early postoperative mortality. Although some methods can avoid the occurrence of PHLF by evaluating liver reserve function, there is still a lack of practical models to accurately predict the risk of PHLF.

In this study, the team screened out four key risk factors, including preoperative Child-Pugh classification, total bilirubin, creatinine,liver resection ratio, by incorporating more than 40 important perioperative characteristics. Five machine learning (ML) algorithms, namely logistic regression (LR), random forest classifier (RFC), extreme gradient boosting (XGB), light gradient boosting machine (LGBM),artificial neural network (ANN), were used to develop an online interpretable dynamic ML model. This model outperforms commonly used clinical scores such as MELD, FIB-4, ALBI,APRI. It can serve as a potential clinical decision support tool to dynamically identify high-risk PHLF patients beforeafter surgery, assisting clinicians in diagnosisearly intervention.

In recent years, the team has carried out scientific research guided by clinical problems,has published three papers in internationally renowned journals. The relevant achievements have been successively incorporatedauthoritative domesticforeign guidelinesconsensus, such as Guidelines for the DiagnosisTreatment of Primary Liver Cancer (2024 Edition)in China, andRecommendations on Robotic Hepato-Pancreato-Biliary Surgery. The Paris Jury-Based Consensus Conference.