Advancing In-Hospital Clinical Deterioration Prediction Models

Abstract

Background Early warning systems lack robust evidence that they improve patients’ outcomes, possibly because of their limitation of predicting binary rather than time-to-event outcomes. Objectives To compare the prediction accuracy of 2 statistical modeling strategies (logistic regression and Cox proportional hazards regression) and 2 machine learning strategies (random forest and random survival forest) for in-hospital cardiopulmonary arrest. Methods Retrospective cohort study with prediction model development from deidentified electronic health records at an urban academic medical center. Results The classification models (logistic regression and random forest) had statistical recall and precision similar to or greater than those of the time-to-event models (Cox proportional hazards regression and random survival forest). However, the time-to-event models provided predictions that could potentially better indicate to clinicians whether and when a patient is likely to experience cardiopulmonary arrest. Conclusions As early warning scoring systems are refined, they must use the best analytical methods that both model the underlying phenomenon and provide an understandable prediction. (American Journal of Critical Care. 2018; 27:381–391)

Value of laboratory results in addition to vital signs in a machine learning algorithm to predict in-hospital cardiac arrest: A single-center retrospective cohort study

“…We chose a random forest model, which is a nonparametric machine learning approach that has been shown to outperform other algorithms without the need for standardization or log-transformation of the input data [ 16 , 21 , 22 ]. We chose this model for the following reasons.…”

“…Second, to obtain a straightforward interpretation of the prediction, we used a classification model rather than a time-to-event model. Classification models were shown to outperform time-to-event models in predicting hospital deterioration in [22].…”

Validation of an Automatic Tagging System for Identifying Respiratory and Hemodynamic Deterioration Events in the Intensive Care Unit

“…In many of these studies the predicted outcome was a defined event that could be easily detected, such as mortality, cardiac arrest, or ICU transfer. In other cases, the predicted end-point was a more complex condition, such as renal failure or sepsis, that required a rule-based algorithm for event tagging [16][17][18][19][20][21]. In most studies, although the rules for event tagging were specified, no details as to the reliability of the tagging, manual or automatic, were provided.…”

Machine learning techniques for mortality prediction in emergency departments: a systematic review

“…4 5 The average rate of mortality among in-patient admissions is around 2% in USA, 6 indicating the necessity of using clinical deterioration prediction models. 7 Moreover, a recent study has shown that some in-hospital deaths due to clinical deterioration can be avoided. 8 Vital signs including heart rate (HR), pulse rate (PR), respiratory rate (RR), oxygen saturation (SpO 2 ), blood pressure (BP), body temperature (TP) and Glasgow coma scale (GCS) are essential parts of the patients' monitoring process.…”

“… 4 5 The average rate of mortality among in-patient admissions is around 2% in USA, 6 indicating the necessity of using clinical deterioration prediction models. 7 Moreover, a recent study has shown that some in-hospital deaths due to clinical deterioration can be avoided. 8 …”