JMIR Publications

ABSTRACT

Background:

It is difficult to predict end-stage liver disease (ESLD) patients that require either acute care or palliative care. In clinical, patients with ESLD have limited treatment options and have a deteriorated quality of life with uncertain prognosis. Early identification of ESLD patients with poor prognosis is valuable especially for palliative care.

Objective:

We sought to create a machine learning monitoring system that can predict mortality or classify ESLD patients. Several machine learning models with visualized graphs, decision trees, ensemble learning, and clustering will be utilized.

Methods:

A retrospective cohort study was conducted using electronic medical records of patients from Wan Fang Hospital and Taipei Medical University Hospital. 1214 patients from Wan Fang Hospital were used to establish a dataset for training, and 689 patients from Taipei Medical University Hospital were used for validation.

Results:

The overall mortality rate of patients in the training set and validation set was 28% (n = 257) and 22.5% (n =145), respectively. In traditional clinical scoring models, PT-INR, which was significant in the cox regression (p-value of <0.001, hazard ratio of 1.288) had prominent influences on predicting mortality. However, the receiver operating characteristic (ROC) curves reached approximately, 0.75. In supervised machine learning models, the ROC curves reached 0.852 for the random forest model, and the ROC curves reached 0.833 for the Adaptive Boosting model. Blood urea nitrogen, bilirubin and Sodium were regarded as the primary factors for predicting mortality. Creatinine, hemoglobin and albumin were also significant mortality predictors. In unsupervised learning, hierarchical clustering analysis can accurately group acute death patients and palliative care patients into different clusters than patients in the survival group.

Conclusions:

Medical artificial intelligence has become a cutting-edge tool in clinical medicine, as it has been found to have predictive ability in several diseases. The machine learning monitoring system in this study contains multifaceted analyses, which provides various aspects in evaluation and diagnosis. This strength makes the clinical results more objective and reliable. Moreover, the visualized interface in this system gives more intelligible outcomes. Therefore, this machine learning monitoring system provided a comprehensive approach for assessing the patients’ conditions, and it may help to classify acute death patients and palliative care patients. Upon further validation and improvement, the system may be used to help physicians in the management of ESLD patients.