Novel Machine-Learning Model Predicts Pneumonia After Liver Transplant

Researchers have developed a novel 14-variable machine-learning model that may be able to predict the risk of pneumonia following after orthotopic liver transplantation. Findings from their study were published in Respiratory Research.

The machine learning model was developed using data from 591 adult patients who underwent orthotopic liver transplantation at a hospital in China between 2015 and 2019. Data were retrospectively extracted from electronic medical records and randomly allocated to training and testing sets. A total of 6 machine learning models were developed with the training set. The developed models included logistic regression (LR), support vector machine (SVM), random forest (RF), adaptive boosting (AdaBoost), extreme gradient boosting (XGBoost), and gradient boosting machine (GBM). The investigators assessed these models with the testing set by the area under curve (AUC) of receiver operating characteristic. Additionally, the investigators also examined any related risk factors and pneumonia outcomes.

Approximately 42.81% (n=253) of patients experienced postoperative pneumonia. Patients who had postoperative pneumonia had a significantly increased risk of hospitalization and death (P <.05). The XGBoost model performed better than the other 6 models in terms of its predictive ability. Using the testing set, the investigators found that the AUC of the XGBoost model was 0.794, with a sensitivity of 52.6% and a specificity of 77.5%.

In the XGBoost model, the researchers observed 14 variables most associated with pneumonia, including the preoperative international normalized ratio, hematocrit, platelets, albumin, alanine transaminase, fibrinogen, white blood cell count, prothrombin time, serum sodium, total bilirubin, anesthesia time, preoperative length of stay, total fluid transfusion, and operation time.

Limitations of the study include its retrospective nature as well as the reliance on patient data from a single center.

The investigators concluded that the XGboost model “holds promise for future clinical application to predict post-transplant pneumonia in [orthotopic liver transplantation] patients.”

Reference

Chen C, Yang D, Gao S, et al. Development and performance assessment of novel machine learning models to predict pneumonia after liver transplantation. Respir Res. 2021;22(1):94. Published online March 31, 2021. doi:10.1186/s12931-021-01690-3