JMIR Publications

ABSTRACT

Background:

The first-year survival among hemodialysis (HD) patients remains poor. The current mortality risk scores for HD patients employ regression techniques and have limited applicability and robustness.

Objective:

We aimed to develop a machine learning model utilizing clinical factors to predict first-year mortality in HD patients that could assist physicians in classifying high-risk patients.

Methods:

Training and testing cohorts consisted of 5351 incident HD patients from a single center and 5828 incident HD patients from 97 renal centers, respectively. The outcome was all-cause mortality during the first year of dialysis. eXtreme Gradient Boosting (XGBoost) was used for algorithm training and validation. Two models were established based on the data obtained at dialysis initiation (model 1) and data from 0-3 months after dialysis initiation (model 2). Ten-fold cross-validation was applied to each model. The C statistic (area under the receiver operating characteristic curve, AUC) was used to assess the predictive ability of the models.

Results:

In the training and testing cohorts, 585 (10.93%) and 764 (13.11%) patients, respectively, died during the first-year follow-up. Fifteen most important features were selected from 42 candidate features. The performance of model 1 (C statistic of 0.8318, 95% CI 0.7771 - 0.8374) was similar to that of model 2 (C statistic of 0.8530, 95% CI 0.8057 - 0.8593).

Conclusions:

We developed and validated two machine learning models to predict first-year mortality in HD patients. Both of the two models could be used to stratify high-risk patients at the early stages of dialysis.