Table 4 Data fusion critical aspects
From: Systems and precision medicine approaches to diabetes heterogeneity: a Big Data perspective
Balancing information from different sources or origins |
Managing conflicting, contradicting and inconsistent data |
Handling missing values |
Differentiating between hard and soft data links, i.e. considering the random processes from which the data are generated as subject to the same parameters, or instead accounting just for covariations, dependencies, similarity/dissimilarity etc. |
Establishing loss or objective functions and regularization/penalty terms |