Modelling hormonal rhythms for biomedical control

Much of the body’s biochemistry changes level according to time of day, season or other variables and there is a great deal of interest in understanding these variations particularly from the perspective of identifying and treating disease. Unlike a "normal", engineering signal processing problem, the available data is typically short (rarely more than one cycle), highly noisy owing to the difficulty in measuring hormone levels in the blood, and often has one or more missing samples. Luckily there are often multiple samples from a number of individuals. All these demand new techniques to model them and draw appropriate inferences. These might include multilevel Bayesian modelling, non-linear (Kalman) filtering or non-linear system identification.

Once a good model is found it can then be used to improve therapy e.g. within open or closed loop, non-linear control. The specific project would be determined in discussion depending on your interests and experience.

You should have a good degree in a relevant, numerate subject with strengths in mathematical analysis, probability & statistics, control theory and strong programming skills e.g. Matlab, Python, C or Java. Experience/training in biology or bio-medicine would be advantageous but is not necessary.