AUTHOR=Chen Wei , Li Xiangkui , Ma Lu , Li Dong 

TITLE=Enhancing Robustness of Machine Learning Integration With Routine Laboratory Blood Tests to Predict Inpatient Mortality After Intracerebral Hemorrhage

JOURNAL=Frontiers in Neurology

VOLUME=Volume 12 - 2021

YEAR=2022

URL=https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2021.790682

DOI=10.3389/fneur.2021.790682

ISSN=1664-2295

ABSTRACT=Objective: The accurate evaluation of outcomes at a personalized level in patients with intracerebral hemorrhage (ICH) is of important clinical implications. In this study, we aim to evaluate the machine learning integrates with routine laboratory tests and electronic health records (EHRs) data to predict inpatient mortality after ICH. 

Methods: In this machine learning-based prognostic study, we obtained 1835 consecutive patients with acute ICH between October 2010 and December 2018. The model building process incorporated 5 pre-implant ICH score variables (clinical features) and 13 out of 59 available routine laboratory parameters. According to a range of learning metrics, we assessed model performance, such as the mean area under the receiver operating characteristic curve [AUROC]. We also used the Shapley additive explanation algorithm to explain the prediction model. 

Results: Machine learning models using laboratory data achieved AUROCs of 0.71 to 0.82 in a split-by-year development/testing scheme, with the nonlinear eXtreme Gradient Boosting model yielded the highest prediction accuracy. In the held-out validation set of development cohort, the predictive model using comprehensive clinical and laboratory parameters outperformed those using clinical alone in predicting in-hospital mortality (AUROC [95% bootstrap confidence interval], 0.899 [0.897 to 0.901] vs 0.875 [0.872 to 0.877]; p < 0.001), with over 81% accuracy, sensitivity, and specificity. We observed similar performance in the testing set. 

Conclusions: Machine learning integrates with routine laboratory tests and EHRs could significantly promote the accuracy of inpatient ICH mortality prediction. These multidimensional composite prediction strategies can become intelligent assistive predictions for ICH risk reclassification and offered an example for precision medicine.