Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.718390
Title: Transforming intensive care data to inform clinical insights : using predictive analytics for traumatic brain injury patients
Author: Mulcahy, Susan
ISNI:       0000 0004 6346 8962
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Abstract:
Traumatic brain injury (TBI) patients in the intensive care unit (ICU) are monitored closely and continuously over several days, yet many of these patients develop secondary insults in the brain. Because the patient is either unconscious or sedated, the patient's condition may be worsening without other measurable signs of a failing condition and the prognosis is elusive. This thesis arises from the viewpoint that examining physiological signals recorded from the patient at high time resolution could give earlier insight into the patient's condition. This research explores heart rate variability, derived from arterial blood pressure signals, using the standard deviation of 5-minute blocks (SDNN5) of heartbeat intervals in the time domain. Significant challenges were met in obtaining real-time high resolution clinical data suitable for computational analysis. Many approaches were attempted including a direct connection through MATLAB, using the Philips Data Export Interface Programming Guide, and third-party softwares Rugloop and ixTrend. The latter, ixTrend, was found to be the most effective. The next step was to establish a data pipeline to import the data in the correct digital format and to organise the multitude of large clinical datasets into usable files (up to 21 million rows of data per file). A binary statistical classifier was developed in MATLAB to predict ICU outcomes of TBI (survivors and nonsurvivors) using supervised machine learning. This measure, SDNN5, is then used to investigate the effect of various standard interventions commonly found in the ICU, where quantifiable measures have previously been lacking. The results for therapeutic hypothermia showed no correlation but a clear effect was seen from endotracheal suctioning. Diurnal variations seem to be overridden by the clinical schedule of medication administration. Lastly, no correlation is seen between spreading depolarisations in the brain (measured by electrocorticography (ECoG)) and SDNN5.
Supervisor: Boutelle, Martyn Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.718390  DOI: Not available
Share: