The increasing volume of available data in the Healthcare domain suggests to use powerful machine leaning techniques for classification and prediction. 

Multiple Sclerosis Progression Analysis

The aim of our work is to leverage the Historical Medical Data of a Patient, also through the ubiquitous wearable devices, to perform tasks as Progression of Disease Prediction, Patient Grouping, Anomaly Detection and, in general, the Support to Diagnosis and Assessment of Life Quality.

In our ongoing recent work, we are studying the Progression of Multiple Sclerosis. This task is complex because the temporal distance between two successive events (e.g. visits) could vary from days to years producing a highly irregular sampled signal. For this reason, we are investigating a novel neural forecasting architecture based on Time-Aware LSTM (Baytas et al. 2017) where the time elapsed between two successive events is opportunely taken in account.

• Lus G., Palmieri F.A.N et al., "Recurrent Neural Network for the prediction of Multiple Sclerosis’ Progression," work in progress