Mads A. Hansen’s work on deep learning for data-driven health involves the development of a so-called deep anchor method, combining the anchor methods for semi-supervised analysis of text data from electronic health records with feature extraction using deep auto-encoders. Mads … Continued
New machine learning paper by Karl Øyvind Mikalsen and co-workers on health analytics! Title: Using Anchors from Free Text in Electronic Health Records to Diagnose Postoperative Delirium Journal: Computer Methods and Programs in Biomedicine Authors: Karl Øyvind Mikalsen, Cristina Soguero … Continued
Intern Andreas Storvik Strauman worked on the project deep learning for data-driven health by developing time series classification tools by the utilisation of recurrent neural networks under the supervision of Filippo Bianchi, Karl Øyvind Mikalsen, Michael Kampffmeyer, and Robert Jenssen. … Continued
Missing data A physicist may say that the world is an approximation of his equations. In the same way, many machine learning models require that the data is “ideal”. With the real world data that is available today, this is not … Continued
Colorectal cancer is one of the highest causes of death worldwide and is estimated to be the most expensive cancer to treat in Norway. But all is not gloomy. Early detection has shown to increase survival rate drastically, which has promted a suggestion for a organized screening program in Norway. But the waiting line for examinations is already long and more screenings would increase the workload on physicians. Automatic screening tools to aid physicians powered by deep learning provides an enticing suggestion.
Kristoffer Wickstrøm joins deep learning for data-driven health as a summer intern! He will work on polyp segmentation using fully convolutional neural networks. Polyps in the intestines may evolve into cancer, and it is important to detect such structures. … Continued