JMIR Publications

ABSTRACT

Background:

More than 20% of patients admitted to the intensive care unit (ICU) develop an adverse event (AE) increasing the risk of further complications and mortality. Despite substantial research on AE prediction, no previous study has leveraged patients’ temporal data to extract features using their structural temporal patterns, i.e. trends.

Objective:

To improve AE prediction methods by using structural temporal pattern detection for patients admitted to the ICU by extracting features from their temporal pattern data to capture global and local temporal trends and to demonstrate these improvements in the detection of Acute Kidney Injury (AKI).

Methods:

Using the MIMIC dataset, we extracted both global and local trends using structural pattern detection methods to predict AKI. Classifiers were built using state-of-the-art models; the optimal classifier was selected for comparisons with previous approaches. The classifier with structural pattern detection features was compared with two baseline classifiers that used different temporal feature extraction approaches: (i) symbolic temporal pattern detection, which is the most common approach for multivariate time series classification; and (ii) the last recorded value before the prediction point, which is the most common approach to extract temporal data in the AKI prediction literature. Moreover, we assessed the individual contribution of global and local trends. Classifier performance was measured in terms of accuracy (primary outcome), area under the curve (AUC), and F-measure.

Results:

Random Forest was the best classifier using structural temporal pattern detection. The accuracy of the classifier with local and global trend features was significantly higher than when using symbolic temporal pattern detection and last recorded value (81.3% vs. 70.6% vs. 58.1%; p<0.001). Excluding local or global features reduced the accuracy to 74.4% or 78.1% respectively (p<0.001).

Conclusions:

We demonstrated that using features obtained from structural temporal pattern detection the onset of AKI prediction in ICU patients was predicted significantly better than with previous approaches. The proposed method of combining both local and global trends is a generalizable approach to predict adverse events in critical care that may be used to help clinicians intervene in a timely manner to prevent or mitigate adverse events.