Exploring the feasibility of vital signs-based mortality risk prediction in a care facility setting

Mortality risk prediction (MRP) enhances healthcare resource allocations and end-of-life care. High prediction accuracy has been reported for MRP in intensive care units (ICUs). However, there have been fewer studies on the use of MRP in care facilities, which lack laboratory test data and continuous vital signs monitoring. Among related studies, vital signs-based therapy (VSbT), common in hospitals and care facilities, can dramatically affect vital signs, but VSbT effect has not been considered in related studies. This lack of consideration might cause lower prediction accuracy.

The purpose of this study was to explore the feasibility of using MRP in care facilities, but with sparse vital signs measurements from nurseChart, part of an open ICU database. To make clear the effect of VSbT on MRP, the authors proposed a feature-exploring algorithm for identifying the VSbT-related features and thereby identify a classifier for vital signs-based MRP. Moreover, appropriate vital signs measurement intervals were investigated using the data of continuous vital signs contained in the database as reference.

This study shows that of all the vital signs, temperature is strongly subject to VSbT effects. Moreover, with sparse vital signs data and certain personal information, the classifier with the proposed VSbT-related features could outperform those reported so far (G-mean: 0.6462 vs. 0.6307). Moreover, for each vital sign, the appropriate measurement interval was determined for care facility scenarios.

Using sparse data from an open ICU database, this study shows the feasibility of vital sign-based MRP use at care facilities, which is a big step towards the practical use of the MRP in those facilities.