Risk Evaluation and Forecasting Causal Relationships

A computational engine is disclosed, which is enabled to discover, learn and forecast causal relationships between a large number of known and hidden variables that evolve over time. The engine is equipped to keep track of a large set of heterogeneous data reported by spatially distributed observers in discrete time intervals. The engine is further suited to be applied to systems where anomalies or outliers have a notable impact on the statistical distribution of variables and where no analytical description for the statistical distribution of the data is known or is difficult to define. Applications range from improved anomaly detection in networks to personalized risk prediction for patients.