Prediction of patient's health risk in critical care using a deep neural network

No Thumbnail Available
Swetha, Jamadaguntla
Issue Date
MSc in Data Analytics
Dublin Business School
Items in eSource are protected by copyright. Previously published items are made available in accordance with the copyright policy of the publisher/copyright holder.
The Intensive care units (ICU’s) of a hospital comprise a large share of the health care budget as today’s lifestyle habits and environmental conditions contribute to the onset of chronic diseases. High risk patients in ICU require extensive monitoring and direct attention from healthcare providers. Improving the quality, efficiency, and effectiveness of healthcare provision is an issue of huge importance. Recent technological advances in machine learning have resulted in innovative solutions for the healthcare industry. This research evaluates state of the art deep learning against traditional machine learning algorithms for predicting patients most at risk in critical care. The proposed deep neural network classifier outperforms the traditional methods such as Logistic Regression, GLM, Naïve Bayes, Random Forest and Decision Tree in terms of accuracy, precision, recall, specificity, AUC, and training time. The best performing state of the art deep learning model has an accuracy of 97%.