Emergency Department Return Prediction System Using Blood Samples With LightGBM for Smart Health Care Services

Abstract

This article proposes a novel Blood sample-based Emergency department (ED) Return (BER) scheme that predicts the ED return probability using LightGBM. In the proposed BER scheme, LightGBM makes predictions on ED return based on blood samples. Since blood sample analysis is one of the most common medical procedures, the proposed scheme can help to improve ED patient care for hospitals. The proposed BER smart health care system and internet of medical things (IoMT) blockchain network was tested from the ED of the Severance Hospital of Yonsei University, located in Seoul of South Korea. The results show that the proposed BER scheme is superior in predicting ED return visits based on achieving a higher Area Under the Curve of the Receiver Operating Characteristic performance, along with the advantage of using much lesser data and being faster.