Sepsis recognition in the department of intensive newborn therapy using an artificial neural network

Hassanin HatemMohamedAbdelMaksoud

Sepsis is the leading cause of morbidity and mortality worldwide. Early detection of sepsis is important because it allows for timely prescribing of potentially life-saving resuscitation and antimicrobial therapy [1]. Sepsis is a person's unregulated response to an infection that causes life-threatening organ dysfunction. According to statistics, [2] approximately one in three deaths in hospitals are associated with sepsis. Although effective protocols exist for the treatment of sepsis, problems remain with early and reliable detection of the condition [3]. In recent years, the increased adoption of electronic health records (EHRs) in hospitals has spurred the development of machine learning surveillance tools for detecting [4] and predicting [5] sepsis. However, most of the existing published models for predicting sepsis are based either on data from a single hospital [6] or from several hospitals in the same health care system where care processes are largely standardized. However, it is noted that the recognition of sepsis in the neonatal non-intensive care unit is practically not considered in the literature. Sepsis is not limited to the intensive care unit. With advances in technology and data granularity that underlie clinical informatics systems, it is now possible to consider the development and implementation of sepsis alert systems in intensive care units and beyond. The rationale for using electronic sepsis surveillance is ultimately to facilitate timely and error-free treatment through early recognition and decision support. However, numerous barriers prevent the development and implementation of hospital-wide sepsis alert systems [7]. The aim of the presented study is to analyze methods for recognizing sepsis in the neonatal intensive care unit using an artificial neural network by analyzing the available electronic data of a medical record.

Keywords:Sepsis, intensive care, artificial neural network, Convolutional neural network, Autoencoder.