Classifying Process Traces for Stroke Management Quality Assessment: A Deep Learning Approach

Stroke management process trace classification can support quality assessment, since it allows to verify whether better-equipped Stroke Centers actually implement more complete processes, suitable to manage complex patients as well. In this paper, we present an approach to stroke trace classification based on deep learning techniques: in particular, we have tested a traditional architecture, based on Recurrent Neural Networks, as well as novel, more complex ones, which combine recurrent networks with convolutional models. Experimental results have shown the feasibility of the approach, and the superiority of composite architectures, which have led to higher accuracy values.