Pollard, T; (2017) Modelling deterioration of health and predicting mortality using the vital signs of critical care patients. Doctoral thesis , UCL (University College London).

Abstract

Advanced bedside monitors are now commonplace in hospitals, presenting clinicians with increasingly large volumes of data on which to guide treatment. Despite these advances, there remains a heavy reliance on clinical intuition and simplistic alarm systems for diagnosis and care planning. Objectively assessing physiological stability using multiple, dynamic variables presents a challenge and early detection of deterioration can be problematic. This interdisciplinary thesis explores how improvements in data-archiving, networking, and sharing present new opportunities for clinical research and health care. Inspired by an initial investigation to predict mortality in patients with an artificial neural network, we move on to develop a probabilistic, data-fusion model for objectively assessing clinical deterioration and recovery. We demonstrate how this model of “normal” physiology can be used to indicate health trajectory, using novel approaches to define the normal population and to construct the model. Concluding the thesis, we highlight potential for the data-fusion model and suggest directions for future research.

Download activity - last month

Export as CSVXMLJSON

Download activity - last 12 months

Export as JSONCSVXML

Downloads by country - last 12 months

Export as JSONCSVXML

Archive Staff Only

View Item