Application of AI Techniques to Predict Survival in Liver Transplantation : A Review

Abstract

The survival of patients after liver transplantation is a major concern in the field of liver transplantation. Because the request for liver transplantation far outnumbers the total of dead donor organs, organ allocation has become critical. Using the medical data available from the transplant patients, it is possible to analyze the key factors accountable for the continued existence of liver transplant patients and hence predict the survival. Traditionally such prediction scores were calculated using regression model. However in the last few decades many changes have taken place due to the usage of Electronic Health records which stores the medical history of patients. This resulted in the use of AI techniques to predict the endurance of liver transplantation by detection of hidden patterns within large datasets. This report offers a systematic review of studies that used AI techniques to predict transplant survival and comparison to already existing models like Survival Outcome following Liver Transplant -SOFT Score, Model for End Stage Liver Disease -MELD Score, Balance of Risk Model -BAR Score and Donor Risk Index -DRI Score. Among all the AI Techniques used, the ANN model outperforms all other existing model with AUROC=0.90 and better exactness on different datasets.