Artificial Neural Networks to Predict Mortality in critical care patients: An application of supervised machine learning
David Andrew Cook · 2005
contemporaneous, formative computer analysis into the delivery and assessment of patient care, with interests in statistical modelling, data integration and operations research. Outcome prediction in intensive care is a challenging process. It requires accurate synthesis of quality data and application of prior experience to the analysis. To facilitate this process, artificial neural network (ANN) technology is being increasingly used. ANNs are a form of artificial intelligence capable of analysing complex medical data. They are a class of models and learning methods that superficially resemble the interconnecting neuronal architecture of the human brain. “Learning ” occurs with iterative changes in the interrelationships between the “neurones”. There is a considerable amount of activity in the application of ANNs to the medical area. A Medline ® search of “artificial neural networks ” currently picks up a list of over a thousand papers. In a recent review1 of artificial intelligence in the Intensive Care Unit (ICU), only six paragraphs described ANN applications in the ICU. This review will provide more background and detail, and will focus on applications of ANNs in