Applications are invited for an experienced researcher to develop and validate advanced statistical methods for analyzing time-series data from adult neuro-intensive care unit (NICU) patients.
Start date: 1st May 2013
Duration: 2 years
Closing date: 5 p.m. UK time on 10th January 2013
Interviews: early February 2013
You will be a probabilistic machine-learning researcher keen to work on a challenging application area. You will be a self-motivated individual with the ability to take day-to-day responsibility for the progress of the proposed work. There is scope for innovative methodological developments in the modelling framework. The role will require you to collaborate and communicate effectively with project partners from medicine. This post offers the opportunity to work in one of the UK's leading machine-learning groups, applying and extending cutting edge methods in an important application area.
The project will focus on the accurate identification of physiological and artifactual events, and on the utility of cleaning artifact from blood pressure data with respect to clinical practice.
In the first year the Research Associate will focus on methods for inferring physiological and artifactual events from time-series data, including data cleansing, anomaly detection, and inference in probabilistic models. This work will build on that of Quinn, Williams and McIntosh (PAMI, 2009) on Factorial Switching Linear Dynamical Systems applied to Physiological Condition Monitoring. In the second year of the project the models will be validated against live data collected at the NICU in the Southern General Hospital (Glasgow), and development of the models continued in light of the results obtained.
The ability to correctly interpret and quantify arterial hypotension events occurring within a noisy clinical environment will lead to a step change in blood pressure management. It will enable the creation of a clinically practical testing environment for the accurate evaluation of new intensive care management strategies whether as part of a clinical audit or randomised clinical trials.
This two-year project is funded by the Chief Scientist Office (CSO) in Scotland. The PI is Prof Chris Williams (School of Informatics, University of Edinburgh). The co-investigators are Dr Ian Piper (Clinical Physics, Southern General Hospital, Glasgow), and clinicians Prof John Kinsella (University of Glasgow), Prof Peter Andrews (University of Edinburgh), and Mr Laurence Dunn (Southern General Hospital, Glasgow).
Informal enquiries about the position may be directed to Prof Chris Williams ckiw@inf.ed.ac.uk .
To apply, and for a full job specification, visit this link (vacancy 007123).
Chris Williams