My research interests are in the statistical analysis of medical data using pattern recognition techniques.
I was awarded a PhD in Statistics by the School of Mathematics and Statistics, University of Glasgow, Scotland in 2014.
The title of my thesis is …
Predicting Hypotensive Episodes in the Traumatic Brain Injury Domain
Download my thesis from this link (PDF)[10.1 MB]
Download a summary of my thesis from this link (PDF)[285 KB]
The data for my research comes from the BrainIT consortium (www.brainit.org.uk)
which has allowed access to the BrainIT database. This extensive resource was collected by the BrainIT organisation during an EU funded project from 2002 to 2005. The database
was the result of a multi-centre study across 22 hospitals in Europe and it contains 3 classes of information: demographic data — providing information about the patient such as age, gender etc.; physiological data — minute by minute readings of vital signs; and episodic data — measurements on any treatments given to the patient, laboratory results and regular Glasgow Coma Scores (GCS) assessments of clinical status.
The BrainIT database contains 199 patients although only 119 patients meet my criteria of having both BPs and BPm available with sufficient low number of short data gaps. The average age of these eligible 119 patients was 39.04 years (range = 15 — 83, median = 34.05), with 85% being male. 19 patients had no events but these patients are included in the training data to reflect the small but present chance of this subgroup. Further details of the BrainIT database are available in:
Piper I, Chambers I, Citerio G, Enblad P,
Gregson B, Howells T, Kiening K,Mattern J, Nilsson P, Ragauskas
A, Sahuquillo J, Donald R, Sinnott R, Stell A (2010) The brain monitoring with information technology
(BrainIT) collaborative network: EC feasibility study results and future direction. Acta Neurochirurgica
http://dx.doi.org/10.1007/s00701-010-0719-1 DOI 10.1007/s00701-010-0719-1
The EU research project Avert-IT has now finished. The project’s research was trying to use advanced pattern recognition techniques to spot hypotensive events (dangerously low blood pressure) in traumatically brain injured (TBI) patients being looked after in an ICU.
The project ran from January 2008 to October 2011. A Phase I clinical study was carried out to assess the infrastructure used to collect clinical data and the ability of the Bayesian Artifical Neural Network (BANN) predicitive engine. Good results allowed the project to move to Phase II clinical study which recruited 49 patients however, due to data collection difficulties, only 39 patient data sets were suitable for analysis.
Here is a link to a presentation on the Phase I results I gave at a conference (ICP 2010) in Tubingen, Germany in September 2010. PDF[838 KB]
Five papers have been published on details of the project. It is intended to publish further papers on the Phase II results.
Anthony Stell, Richard Sinnott, Jipu Jiang, Rob Donald, Iain Chambers, Giuseppe Citerio, Per Enblad, Barbara Gregson, Tim Howells, Karl Kiening, Pelle Nilsson, Arminas Ragauskas, Juan Sahuquillo, and Ian Piper. Federating distributed clinical data for the prediction of adverse hypotensive events. Philosophical Transactions of the Royal Society A: Mathematical,Physical and Engineering Sciences, 367(1898):2679–2690, 2009. doi: 10.1098/rsta.2009.0042 URL http://rsta.royalsocietypublishing.org/content/367/1898/2679.abstract
A. Stell, R.O. Sinnott, R. Donald, I. Chambers, G. Citerio, P. Enblad, B. Gregson, T. Howells, K. Kien- ing, P. Nilsson, A. Ragauskas, J. Sahuquillo, and I. Piper. A distributed clinical data platform for physiological studies in the brain trauma domain. Sixth IEEE International Conference on eScience, 1:65 – 72, 2010. doi: 10.1109/eScience.2010.26 URL http://dx.doi.org/10.1109/eScience.2010.26
Rob Donald, Tim Howells, Ian Piper, I. Chambers, G. Citerio, P. Enblad, B. Gregson, K. Kiening, J. Mattern, P. Nilsson, A. Ragauskas, Juan Sahuquillo, R. Sinnott, and A. Stell. Early warning of EUSIG-defined hypotensive events using a bayesian artificial neural network. In Martin U. Schuhmann and Marek Czosnyka, editors, Intracranial Pressure and Brain Monitoring XIV, volume 114 of Acta Neurochirurgica Supplementum, pages 39–44. Springer Vienna, 2012b. ISBN 978-3-7091-0956-4. doi: 10.1007/978-3-7091-0956-4_8 URL http://dx.doi.org/10.1007/978-3-7091-0956-4_8
Rob Donald, Tim Howells, Ian Piper, I. Chambers, G. Citerio, P. Enblad, B. Gregson, K. Kiening, J. Mattern, P. Nilsson, A. Ragauskas, Juan Sahuquillo, R. Sinnott, and A. Stell. Trigger char- acteristics of EUSIG-defined hypotensive events. In Martin U. Schuhmann and Marek Czosnyka, editors, Intracranial Pressure and Brain Monitoring XIV, volume 114 of Acta Neurochirur- gica Supplementum, pages 45–49. Springer Vienna, 2012a. ISBN 978-3-7091-0956-4. doi: 10.1007/978-3-7091-0956-4_9 URL http://dx.doi.org/10.1007/978-3-7091-0956-4_9
Anthony Stell, Richard Sinnott, Rob Donald, and Ian Piper. Supporting clinical trials to predict adverse events in the brain trauma domain. In The 25th IEEE International Symposium on Computer-Based Medical Systems (CBMS 2012), 2012.