Deep Learning is a new approach of machine learning that set new standards in multiple fields over the last five years. As the name suggests, the models have deeper and hierarchical structures. The models, loosely based on biological neural networks, can due to this structure work on a more abstract level and can reduce the complexity of the data accordingly.
The hierarchical structure replaces many preprocessing steps, which are now learned as part of the model. In the past, methods had to be developed in laborious engineering work over many years to reduce data complexity. The field of application is very diverse and the potential is great. In Switzerland, however, this potential is far from being exhausted.
In our product Casematch, the Deep Learning approach is already in use. The program accesses a data pool to acquire the knowledge contained in the data in a self-learning process and then uses it for the DRG controlling. The implicit coding knowledge from routine data is utilized and a knowledge transfer between different coding practices takes place in order to increase the coding accuracy.
Deep Learning has only developed in the last few years and is characterized by new trends and innovations. We meet this challenge with an unique combination of knowledge and the ambition to always be one step ahead.
Read our blog post (only in German) for more information.