Comparative Study Of Artificial Intelligence Model And ICG In Evaluating Liver Reserve Function Of Patients With Hepatocellular Carcinoma

Objective:The aim of the present study was to construct an artificial intelligence(AI)model to predict preoperative indocyanine green retention rate at 15 minutes(ICG-R15)in patients with hepatocellular carcinoma(HCC)based on clinical data,and to explore its prognostic value.Methods:The data of the HCC patients(N=565 cases)who underwent hepatectomy at Sichuan Provincial People’s Hospital from September 2015 to September 2022 were retrospectively analyzed.Quantitative data of normal distribution were expressed as x±s,quantitative data of skewed distribution were expressed as M(IQR),and qualitative data were expressed as absolute numbers or percentages.The AI model was constructed using the Light Gradient Boosting Machine(Light GBM)method in Python platform.The prediction performance was evaluated by using the linear regression equation correlation coefficient R~2and the Receiver Operating Characteristic curve(ROC curve).Results:A total of 469 males and 96 females,aged 56(±12)years,were included in this study.Among them,452 patients were included in the model training set and 113patients were included in the validation set.The AI model for predicting ICG-R15 was successfully constructed in this study.The linear fit between the Estimated Indocyanine Green 15 min Retention and ICG-R15 showed good results(R~2=0.94).eICG-R15<20.0%plotted ICG-R15<19.0%of subjects with the area under ROC curve(AUC)of 0.984(95%CI:0.973-0.996)achieved satisfactory outcome in predicting ICG-R15.The AUCs of eICG-R15 and ICG-R15 for predicting postoperative liver failure(PHLF)were 0.749(95%CI:0.687-0.811)and 0.755(95%CI:0.694-0.815),respectively,indicating that ICG-R15 and eICG-R15 had similar predictive efficacy in assessing PHLF.Conclusion:AI model can predict theICG-R15 value of HCC patients individually,so as to replace the indocyanine green clearance experiment to a certain extent.At the same time,it can also help primary medical institutions with limited economic level and lack of professionals to evaluate the liver reserve function before surgery,formulate surgical plans,and avoid the risk of reagent sensitization and data errors caused by instruments and operations.