Modified Early Warning Score (MEWS) Visualization and Pattern Matching Imputation in Remote Patient Monitoring

Abstract

Remote Patient Monitoring (RPM), which leverages the Internet of Medical Things (IoMT) and autonomous systems, has grown in popularity recently. In RPM, the IoMT sense a patient’s biophysical data and transmits it in real time while the autonomous system processes the data for clinical notifications and storage. However, RPM deployments face two diverse challenges: how to present continuous data so that healthcare professionals can quickly interpret data streams and how to manage a great deal of missing data that occurs in RPM. Several studies suggested techniques for imputing missing data in static databases, which are unsuitable for RPM. A method for constantly streaming healthcare data to medical experts involves summarizing vital signs information into a numerical score, such as the Modified Early Warning Score (MEWS), which may be visually displayed to highlight MEWS patterns over a certain period. However, a MEWS chart is simplistic and more sophisticated ways to present data visually for straightforward interpretation are needed. This research proposes a solution for the visualization and missing data challenges by identifying patterns in the RPM data. First, a pattern-matching technique is proposed to address the missing data by considering the correlation and variability of the vital signs, resulting in a comparable correct match rate. Second, we transform the observed raw physiological vital signs data into concepts we call trust, frequency, trend, and slope parameters for visualization and automated alerts. The proposed approach can better support clinical decision-making than the MEWS. Comprehensive visualization approaches and missing data solutions can improve the quality and dependability of patient risk assessments.

Keywords

• Pattern imputation
• data integrity
• visualized remote patient monitoring (VRPM)
• clinical decision support
• wearable technology