Projects

This project studies the link between severe asthma attacks and local weather conditions using a mobile telephone telemedicine system. The results suggest that individual tailor-made management plans that take account of local weather conditions could be of added value for monitoring and self-management of asthma. More details.

A mobile phone telemedicine system has been developed to assist patients in the self-management of Type 1 diabetes. The system has been evaluated in a clinical trial, and the dataset gathered during the trial is being used to develop algorithms to target nurse support more effectively, and provide better feedback to patients. More details.

ECGSYN is a model based on three coupled ODEs which generates a synthesised ECG signal with user-settable mean heart rate, number of beats, sampling frequency, waveform morphology, RR interval, and LF/HF ratio. It reproduces many of the typical features of the human ECG. Available in Matlab, Octave, C and Java implementations, the output may be employed, for example, to assess signal processing techniques used to compute clinical statistics from the ECG. More details.

This project investigates the analysis of a sequence of temperature fields obtained from a pool nucleate boiling experiment. A new method developed using non-orthogonal empirical functions (NEFs) overperforms Principal Component Analysis (PCA) by encoding the relevant physical constraints, providing an efficient identification of the non-stationary positions of active nucleation sites in successive frames. More details.

Epilepsy is one of the most common serious neurological disorders, affecting 1% of the population at some stage. The ability to detect and predict epileptic seizures would facilitate clinicians and may provide a means of improving treatment for many sufferers. We analyse techniques based on traditional linear statistics and nonlinear methods which can be applied to both scalp and intracranial electroencephalogram (EEG) signals. More details.

Recent observations of sleep-wake transitions observed during sleep have been incorporated into a new model of beat-to-beat heart rate variations over the course of 24 hours. This new model faithfully represents some of the key empirical characteristics that have been observed in actual recordings of healthy human subjects, for example the existence of scale-free power law distributions during wakefulness and the exponential distribution of sleep-wake activity during prolonged sleep. This model can be used to assess biomedical signal processing methods applied to the RR interval time series. More details.

Deregulation of electricity markets has increased the need for reliable forecasts of demand, supply and spot price. Forecast horizons, ranging from minutes to several years, are required for ensuring system stability, maintenance scheduling and capital planning. A multidisciplinary approach fusing economics, mathematical modelling and statistics is proposed to combine weather forecasts, economic indicators, operating characteristics of the generators, a model of the micro-market and a deterministic model of seasonality and holidays. The project's objective is the generation of reliable probabilistic forecasts which are essential for making decisions and managing risk. More details.

We have developed novel nonlinear models for speech compression, biometrics and morphing applications. Additionally, we have developed a very sensitive statistical test that detects hidden nonlinear dynamics in speech inaccessible to standard techniques, allowing us to very accurately pick out the correlations due to nonlinear dynamics. It is likely that these dynamics help differentiate one individual from another, which could be used as a reliable biometric marker. More details.

The dynamics of disordered speech is highly complex, and existing techniques for automated screening do not take account of these dynamics correctly. This project has produced a simple method for analysing the complexity of biophysical oscillation and extent of turbulence in speech pathologies, resulting in a method suitable for computationally simple, remote, clinical screening. More details.

Software packages allow policy makers to easily adapt, implement and explore system dynamics models of complex interactions between numerous variables, representing different policy scenarios. This project shows that Genetic Algorithms (GA) provide an efficient and accurate method for identifying optimal choices from among the vast number of possible scenarios that are available, that can be utilised to arrive at better strategies for government policy. More details.

In Frankfurt am Main, on the first Thursday of each month the President of the European Central Bank gives a press conference in order to say whether or not interest rates have been changed, and why. To test the hypothesis that foreign exchange market volatility is affected by such speeches, this project undertakes a textual analysis of the speeches and compares this with movements in the foreign exchange markets. The analysis uses linear correlation, mutual information, a hidden Markov model and a stochastic volatility model driven by a recurrent neural network. More details.

Wind energy is associated with one of the lowest costs of electricity production, the largest resource availability, and a level of technical efficiency presently mature enough to create a substantial contribution. The intermittent nature of wind power generation presents a challenge for wind farm operation and electricity system management. This project uses the probability of exceeding a particular wind speed to create short- to medium-term forecasts of wind power output, critical both for technical, financial and risk management purposes. More details.

Detecting abnormal operating behaviour in jet engines is of critical importance in hazard avoidance. Engine manufacturers are adopting a condition monitoring maintenance approach, with embedded intelligent analysis systems processing data from sensors. This project introduces methods for analysing engine vibration data, providing in-flight abnormality detection, and ground-based engine monitoring using nonlinear techniques. More details.

Devastating floods due to extreme rainfall are predicted by some to increase in severity over the next century. This NERC-funded project, combines, for the first time, mathematical forecasting techniques, supercomputer ensemble numerical weather modelling, and the largest electronic archive of extreme rainfall events to UK flood risk estimation. It uses probabilistic forecasting and machine-learning techniques to predict and categorise devastating extreme rainfall events. More details.

Several methods have been proposed to infer reaction mechanisms for metabolic pathways and networks given abundant time series data. We have developed a data-based principled (DBP) modelling approach which combines first-principles with data-based modelling in an attempt to construct parsimonious models. This provides an automated modelling approach that is tailor-made to biochemical reactions. More details.

In the wake of the FIFA decision to ban high-altitude football, this study investigates the relationship between goal scoring performance of home and away teams of different altitudes. The results suggest that the home-away altitude difference significantly affects the outcome of international football games. More details.

One consequence of an ageing population is the predicted increase in incidence of crippling neurological disorders such as Alzheimer's and Parkinson's disease, leading to increasing demand on health services. The ability to automatically and remotely monitor such disorders is of value in a wide variety of clinical contexts, from pre-screening to the assessment of treatment progress. This telemedicine project seeks to provide simple, remote measures of symptom severity. More details.