In addition to the Tewfik's answer, you might want to take a look at Kalman-Bucy filter. It is interesting to know that Kalman filter, usually used for continuous states (do not confuse with time continuity), can be also utilized for population models like here http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=5531107.

Sorry, I have not studied in this regard, thank you!

Sorry for the kind of advertisement, but I would also like to recommend the paper on the parameter estimation for CTMCs (that come from biological models) using Maximum Likelihood approach. And as Tewfik Kernane said, there is a huge amount of works on the statistical inference of parameters.

http://www.amazon.com/Information-Theory-Inference-Learning-Algorithms/dp/0521642981

http://www.oxfordscholarship.com/view/10.1093/acprof:oso/9780199694587.001.0001/acprof-9780199694587-chapter-23

http://www.cs.toronto.edu/~radford/review.abstract.html

Article Approximate maximum likelihood estimation for stochastic che...