Prediction of Adverse Effects of Medical Treatment using Electronic Health Record and Machine Learning

In the past few years, deaths due to adverse effects of medical treatment have become a major concern and early detection and diagnosis can lead to a timely cure. There is a significant need to develop automated systems for adverse event detection. The complete blood count (CBC) test is a common broad screening test used to determine patients' overall health status. CBC data can be used to screen for a variety of conditions and diseases and also to monitor the efficacy of the treatment. Considering the facts, in this study authors have proposed to use ensemble machine learning methods to develop an automated system for the detection of patients with adverse effects of medical treatment (AEMT) symptoms. The proposed approach has been evaluated on an electronic health record dataset. The data consists of 10 parameters with instances labelled as in-care and out-care. The study results indicate that the extremely randomized ensemble learning methods show better performance as compared to other methods with a reduction in execution time. The developed automated system can be used for real-time medical assistance.