A heterogeneous multi-modal medical data fusion framework supporting hybrid data exploration

Area of Science:

• Medical Informatics
• Data Science
• Healthcare Technology

Background:

• Industry 4.0 drives increased use of high-tech devices in healthcare.
• Medical data is increasingly heterogeneous and multi-modal, including unstructured imaging data (X-rays, MRI, CT, PET).
• Traditional data warehouses struggle with real-time processing, high costs, and handling multi-modal data fusion and exploration.

Purpose of the Study:

• To propose an efficient data fusion framework for heterogeneous multi-modal medical data.
• To overcome limitations of traditional data warehouses in managing and exploring complex medical datasets.
• To enable better data accessibility for healthcare researchers and users.

Main Methods:

• Development of a data lake-based framework for data fusion.
• Implementation of a novel method to fuse fragmented multi-modal medical data and store metadata.
• Creation of a user-friendly interface supporting hybrid graph queries for data exploration.
• Inclusion of indexes to accelerate hybrid data exploration.

Main Results:

• A highly efficient data fusion framework was developed.
• The framework successfully fuses heterogeneous multi-modal medical data.
• A prototype demonstrated effectiveness in a hospital setting.
• Hybrid graph queries enable efficient multi-modal data exploration.

Conclusions:

• The proposed data lake framework offers an effective solution for managing and exploring multi-modal medical data.
• This approach enhances the extraction of diverse information from complex medical datasets.
• The framework improves data accessibility and usability for healthcare research.