Mechanical ventilation is a lifesaving tool and provides organ support for patients with respiratory failure. However, injurious ventilation due to inappropriate delivery of high tidal volume can initiate or potentiate lung injury. This could lead to acute respiratory distress syndrome, longer duration of mechanical ventilation, ventilator associated conditions and finally increased mortality. In this study, we explore the viability and compare machine learning methods to generate personalized predictive alerts indicating violation of the safe tidal volume per ideal body weight (IBW) threshold that is accepted as the upper limit for lung protective ventilation (LPV), prior to application to patients. We process streams of patient respiratory data recorded per minute from ventilators in an intensive care unit and apply several state-of-the-art time series prediction methods to forecast the behavior of the tidal volume metric per patient, 1 hour ahead. Our results show that boosted regression delivers better predictive accuracy than other methods that we investigated and requires relatively short execution times. Long short-term memory neural networks can deliver similar levels of accuracy but only after much longer periods of data acquisition, further extended by several hours computing time to train the algorithm. Utilizing Artificial Intelligence, we have developed a personalized clinical decision support tool that can predict tidal volume behavior within accuracy and compare alerts recorded from a real world system to highlight that our models would have predicted violations 1 hour ahead and can therefore conclude that the algorithms can provide clinical decision support.

No abstract

Computer systems deployed in hospital environments, particularly physiological and biochemical real-time monitoring of patients in an Intensive Care Unit (ICU) environment, routinely collect a large volume of data that can hold very useful information. However, the vast majority are either not stored and lost forever or are stored in digital archives and seldom re-examined. In recent years, there has been extensive work carried out by researchers utilizing Machine Learning (ML) and Artificial Intelligence (AI) techniques on these data streams, to predict and prevent disease states. Such work aims to improve patient outcomes, to decrease mortality rates and decrease hospital stays, and, more generally, to decrease healthcare costs.This chapter reviews the state of the art in that field and reports on our own current research, with practicing clinicians, on improving ventilator weaning protocols and lung protective ventilation, using ML and AI methodologies for decision support, including but not limited to Neural Networks and Decision Trees. The chapter considers both the clinical and Computer Science aspects of the field. In addition, we look to the future and report how physiological data holds clinically important information to aid in decision support in the wider hospital environment.

No abstract

IntroductionElectronic clinical decision support (eCDS) tools are used to assist clinical decision making. Using computer-generated algorithms with evidence-based rule sets, they alert clinicians to events that require attention. eCDS tools generating alerts using nudge principles present clinicians with evidence-based clinical treatment options to guide clinician behaviour without restricting freedom of choice. Although eCDS tools have shown beneficial outcomes, challenges exist with regard to their acceptability most likely related to implementation. Furthermore, the pace of progress in this field has allowed little time to effectively evaluate the experience of the intended user. This scoping review aims to examine the development and implementation strategies, and the impact on the end user of eCDS tools that generate alerts using nudge principles, specifically in the critical care and peri-anaesthetic setting.Methods and analysisThis review will follow the Arksey and O’Malley framework. A search will be conducted of literature published in the last 15 years in MEDLINE, EMBASE, CINAHL, CENTRAL, Web of Science and SAGE databases. Citation screening and data extraction will be performed by two independent reviewers. Extracted data will include context, e-nudge tool type and design features, development, implementation strategies and associated impact on end users.Ethics and disseminationThis scoping review will synthesise published literature therefore ethical approval is not required. Review findings will be published in topic relevant peer-reviewed journals and associated conferences.