A dual boundary classifier for predicting acute hypotensive episodes in critical care

Abstract

An Acute Hypotensive Episode (AHE) is the sudden onset of a sustained period of low blood pressure and is one among the most critical conditions in Intensive Care Units (ICU). Without timely medical care, it can lead to an irreversible organ damage and death. By identifying patients at risk for AHE early, adequate medical intervention can save lives and improve patient outcomes. In this paper, we design a novel dual–boundary classification based approach for identifying patients at risk for AHE. Our algorithm uses only simple summary statistics of past Blood Pressure measurements and can be used in an online environment facilitating real–time updates and prediction. We perform extensive experiments with more than 4,500 patient records and demonstrate that our method outperforms the previous best approaches of AHE prediction. Our method can identify AHE patients two hours in advance of the onset, giving sufficient time for appropriate clinical intervention with nearly 80% sensitivity and at 95% specificity, thus having very few false positives.

Dynamic Prediction of Patient Outcomes in the Intensive Care Unit: A Scoping Review of the State-of-the-Art

“…Two models predicted hypotensive episodes 17,22 . Acute hypotensive episode is a sudden onset of a period of sustained low blood pressure 17 .…”

“…When looking at studies predicting various cardiovascular complications, the prevalence was also very variable: between 9.1% and 35.0%. This is because hypotensive episodes and haemodynamic instability are more common complications, especially in cardiac patients, who were included in Hernandez et al's, Bhattacharya's and Lee's datasets, resulting in high number of patients with the predicted outcomes 17,22,51 69 . This shows that some complications that are reported without specific criteria based on laboratory results or vital signs can be under-reported in the electronic health records.…”

“…It was common to use summary statistics, such a mean or median, minimum and maximum, first and last values within pre-determined time windows 19,20,28,31,33,34,[36][37][38][42][43][44][45][46] . For some models, the classification method handled the time-series data 17,22,27,29,32,39,41,50 . However, some studies used special methods to include the temporal aspect into their model was then summarised, using splines, resulting in coefficients that were used to capture information about the shape of the time series 25 .…”

“…The closest to "real-time" models were those that updated their prediction every time new measurements were entered into the system. Eleven studies followed this prediction frequency 8,17,26,27,36,39,41,42,48,50,51 .…”

Multi-disease Predictive Analytics: A Clinical Knowledge-aware Approach

“…Wide adoption of EHR in recent years has led to the development of many machine learning models for diagnosis prediction using historical EHR data comprising clinical investigations, drug prescriptions, previous diagnoses, and demographic information, e.g., References [11,24,31,33]. Majority of the previous works on diagnosis prediction has focused on the prediction of specific single diseases, e.g., References [3,13,14,55]. Data-driven models for MDPA can be formulated through multi-task learning (MTL) or multi-label learning (MLL).…”