Generalized aggregation index based collaborative fusion for medical diagnosis.

Abstract

It is critical for data fusion and its decision applications to define the similarity relationships between data units. Compared with the traditional similarity relationship driven by data, this paper proposes a new concept of generalized aggregation index (GAI) driven by data and knowledge. The concept of GAI defines the cumulative representation of the generalized relationships between data units. The generalized relationship involves not only data attributes but also the decision effect knowledge behind data attributes, and is a more accurate relationship representation. Based on data units' GAIs, a new GAI based collaborative fusion method for multi-source medical data is proposed to get high quality fusion results and precise decision conclusions. In the fusion process, the data units in different datasets attract each other and aggregate into entity subsets for fusion collaboratively based on the data units' attraction capabilities measured by data units' GAIs. The experimental analysis shows that the proposed classical and quantum-inspired GAI based methods can get high quality fusion results and highly precise diagnosis conclusions.