The winning paper (Passini et al., 2017) describes an in silico model built on simulations of human cardiac cells that was challenged with 62 drugs and reference compounds. The model predicted the risk of drug-induced heart arrhythmia in humans with 89% accuracy in comparison to 75% accuracy found in previously conducted animal trials.
The prize consists of a £28,000 grant and a £2,000 personal award. The competition is open to researchers in academia and industry based anywhere in the world and recognises a paper published in the last three years that has significant impacts on the replacement, reduction or refinement of animals in research. It has been awarded by the NC3Rs since 2005 with sponsorship from GSK.

NC3Rs press release

Reference
Passini, E., Britton, O. J., Lu, H. R. et al. (2017).  Human in silico drug trials demonstrate higher accuracy than animal models in predicting clinical pro-arrhythmic cardiotoxicity.  Frontiers in Physiology 8, 668. doi:10.3389/fphys.2017.00668