Olivia, Diana and Nayak, Ashalatha and Balachandra, Mamatha and John, Jaison (2020) A classification model for prediction of clinical severity level using qSOFA medical scor. Information Discovery and Delivery. ISSN 2398-6247

PDF 8288.pdf - Published Version Restricted to Registered users only Download (1MB) | Request a copy

Abstract

The objective of this study is to develop a multi-label classification model for predicting the future clinical severity level of a patient using observed vital sign values based on the qSOFA medical scoring system. At the ED or ICU, continuous monitoring of a large number of patients and prioritizing them based on their mortality risk level is difficult. Our developed system can make predictions of the future clinical severity level of a patient in real-time using the learned temporal correlations of multiple vital signs and various medical scores. Additionally, using the K-Means technique we claim that the occurrence of a severity risk level change episode depends on the sequence of immediate past data of that event. In addition, we identified the significance of the various medical scores and their correlation in model accuracy. In this paper, we have developed a multi-label classifier using Neural Network and ensemble RAkEL algorithm. The model makes use of basic vital sign information, correlation and statistical features among the vital signs and various medical score systems such as NEWS, MEWS, SIRS, and MODS. The model is trained using formulated training dataset based on data from physiobank archive using qSOFA scoring system. We demonstrated that RAkEL performs better than the Neural network.

Actions (login required)

View Item