At my work we are using them(K20s) to interpret and analyse large EEG data sets, so hopefully someday we will contribute towards some advancements in the treatment of Dementia and Alzheimers disease.
My role so far has been to convert existing MATLAB code to faster CUDA versions. The researchers I work with have been ‘first-names’ on papers related to EEG analysis, but nothing yet directly related to this batch of current 128 channel EEG work.
Soon a new paper on a conic optimization problem solver will be published, and there will be a link to my CUDA version of the solver(in addition to the author’s MATLAB and C version). I will link to that once it is published.