The recent developments in biomedical sensing technologies for ambulatory use have mostly focused on the creation of wearable devices applied to a subjects? body. However, a new paradigm is emerging, which seeks sensor integration in everyday use objects. This new paradigm, dubbed ?invisibles?, is prone to greatly improve the usability of medical devices, since measurements are made as an extension of the natural interaction between the subject and the sensorized objects. One everyday use object that is currently widespread is the smartphone and, due to their prevalence, cardiovascular disorders (CVDs) triage is an appealing target application. Within CVDs, Electrocardiography (ECG) is a first line exam. State-of-the-art hardware technologies allow the acquisition of ECG signals at the hands / fingers, using sensors that can be integrated into a smartphone (e.g.). However there are still multiple bottlenecks associated with this acquisition methods and that can benefit from advanced data science methods. Our aim with this work is to evaluate the performance of AI methods in Atrial Fibrillation detection, using data from the PhysioNet/Computing in Cardiology Challenge 2017 AliveCor dataset.