Predicting Short-term Survival after Liver Transplantation using Machine Learning

Abstract

Liver transplantation is one of the most effective treatments for end-stage liver disease, but the demand for livers is much higher than the available donor livers. Model for End-stage Liver Disease (MELD) score is a commonly used approach to prioritize patients, but previous studies have indicated that MELD score may fail to predict well for the postoperative patients. This work proposes to use data-driven approach to devise a predictive model to predict postoperative survival within 30 days based on patient's preoperative physiological measurement values. We use random forest (RF) to select important features, including clinically used features and new features discovered from physiological measurement values. Moreover, we propose a new imputation method to deal with the problem of missing values and the results show that it outperforms the other alternatives. In the predictive model, we use patients' blood test data within 1-9 days before surgery to construct the model to predict postoperative patients' survival. The experimental results on a real data set indicate that RF outperforms the other alternatives. The experimental results on the temporal validation set show that our proposed model achieves area under the curve (AUC) of 0.771 and specificity of 0.815, showing superior discrimination power in predicting postoperative survival.

Figure 1

The top important features selected from RF with step-wise selection. These variables are verified by the physician. (DX1: reason for liver transplantation, type: type of hepatitis, HCC: Hepatocellular carcinoma).

Figure 2

Experimental results with different range of days as the data source. The AUC increases as more training data are used in the model, and the data of day 1 to day 9 is the most important one.

Figure 3

Hazard ratios (HR) from Cox proportional hazards model with the data of day 9. In the results, HR >1indicates an increased risk of death, and HR < 1 represents a decreased risk. The p-values of the variables show that INR, Platelets and age are significant features.

Figure 4

Experimental flow. The experimental flow comprises data pre-processing, feature selection, imputation of missing values, model training and evaluation. The purpose of training data is for model training, whereas testing data is used for model evaluation.