Discharge prediction of critical patients with spinal cord injury: a machine learning study with 1485 cases

Abstract

Objectives: Prognostication of spinal cord injury (SCI) is vital, especially for critical patients who need intensive care. The study aims to develop machine-learning (ML) classifiers for discharge prediction of SCI patients in the intensive care unit (ICU). Methods: Clinical data of patients diagnosed with SCI were extracted from the publicly available ICU database. A total of 105 ML classifiers were initially developed to predict the discharge destination (dead, further medical care, home), and then the top 3 classifiers with the best performance were stacked into an ensemble classifier (Esb-Clf). To balance the accuracy and the feasibility, the complete Esb-Clf was finally simplified with top 10 features (simplified Esb-Clf). The micro-average area under the curve (AUC) was used to compare the prediction performance of difference ML classifiers and 6 doctors' artificial prediction. Results: A total of 1485 SCI patients were used for the early and the recent prediction of discharge destination. In the early prediction, the micro-average AUC of the Esb-Clf and the simplified Esb-Clf was 0.846 and 0.835 during the independent testing, respectively. While in the recent prediction, the micro-average AUC of the Esb-Clf and the simplified Esb-Clf was 0.898 and 0.892. Performance of both the Esb-Clf and the simplified Esb-Clf were superior to the doctors' in the early and the recent prediction. Conclusions: ML classifiers can discriminate the discharge destination of SCI patients with high accuracy, feasibility and interpretability. Whether the simplified Esb-Clf as an online predictive tool is applicable to guiding clinical management needs further verification.

The Impact of Machine Learning and Robot-Assisted Gait Training on Spinal Cord Injury: A Systematic Review and Meta-Analysis

“…The current ML study [43] demonstrates a higher test accuracy of 73.6% than the MRI accuracy of 71.4%. A few studies [43][44][45][46][47] collectively utilize a patient sample size that is one to two times larger and incorporates a comprehensive evaluation of feature importance. Furthermore, even considering all AIS grades and employing a far less complex model that can be readily implemented, the outcomes are generally similar or superior.…”

Spinal Cord Injury AIS Predictions Using Machine Learning

“…The inclusion of imagery from MRI as a set of features is a possible route of future research that could further bolster the Ridge Classifier as well. In contrast to the research performed in studies by Inoue et al (2020) , Fan et al (2021) , Buri et al (2022) , Chou et al (2022) , and Okimatsu et al (2022) as a whole, the study conducted here uses a patient base one to two times larger, while including a comprehensive review of feature importance as well. To add on, the results are, overall, comparable or better while considering all AIS classes and using a very lightweight model that can be much more easily deployed.…”

“…Researchers have developed many tools on SCI treatment; however, after literature review, it was seen that there is a wide gap in the use of machine learning algorithms to predict SCI recovery in a contemporary precision medicine context, especially with regard to feature importance and using a very large dataset ( Snoek et al, 2004 ; Munce et al, 2014 ). One study attempted to predict discharge location using an ensemble model and used area under the curve as an outcome ( Fan et al, 2021 ), another study made use of convolutional neural nets (CNNs) on MRI charts to achieve an accuracy of 71.4% ( Okimatsu et al, 2022 ), while two other studies greatly limited the complexity to specific AIS scores of A ( Buri et al, 2022 ) and D/E ( Inoue et al, 2020 ). The authors in the study by Chou et al (2022) conducted a study similar to the one presented here, but their sample size consisted of 74 patients.…”