I think that this might be a good place to start:

"Gaze Contingent Displays: A Review"

Duchowski, Cournia & Murphy

Predicting a user's next action is a pretty hard problem which I imagine is most adequately met by performing some user experience research. What kind of interface are you thinking of? If it's a website, maybe tracking navigation data from existing users would be a good jumping off point. 

http://www.ncbi.nlm.nih.gov/pubmed/15687796

Hi Nelson,

We have a software solution that would allow you to build and test rich UIs that would be adaptive to real-time results of eye-tracking analysis. A demo of the analysis is under the link below. If you are interested, contact me directly.

https://www.youtube.com/watch?v=GZnPoAOsXto

Prediction of eye movement of user is hard problem but as far as I know there was an new approach to identify the next action of user while they were in limited world such as games.

I can only say that new approach was presented at ECEM (European Conference on Eye Movements) few days ago, that you may reach the full detail by using the paper which presented by 

Keynote Lecture:

Laurent Itti

University of Southern California, US (Lab Website)

and the name of the paper was

Computational modeling of bottom-up and top-down visual attention in complex dynamic environments

good luck

 And I wanted to add information about the fixation , saccades that generally trackers use algorithm  where the algorithms stands for identifying the saccades and fixation however, these algorithms generally depends on some assumption where some of them use velocity of the eyes some of them use spatial information of gaze data. I think it can be good to developed an paradigm where analysis can be done by using raw data. 

Thank you so much for your contributions until now.

I am focused in data search applications where the data collected from eye tracker can help in inferring the interests of the users. Also, I am exploring new approaches for the development of my PhD thesis. Right now, I am not so interested in using existent commercial solutions, although I am willing to test and review them.

The topic of adaptive user interfaces, prediction of next user's actions and real time analysis of eye tracker data are essential at this time.

Once more, thank you for your contributions, they are very valuable to me.