Graph-based visualization of sensitive medical data
Ilias Kalamaras,Konstantinos Glykos,Vasilis Megalooikonomou,Konstantinos Votis,Dimitrios Tzovaras
DOI: https://doi.org/10.1007/s11042-021-10990-1
IF: 2.577
2021-06-25
Multimedia Tools and Applications
Abstract:With the increasing amounts of electronic health data being constantly generated in medical examinations and by sensors and mobile applications, data visualization methods can assist medical professionals and researchers in exploring and making sense of the data. Two important challenges faced by data visualization are large data volume and protection of sensitive data. In this paper, we propose a graph-based method that allows the exploration of a patient dataset, while also naturally allowing the summarization of large amounts of data, making it applicable to large datasets and sensitive data. A graph is constructed from the raw data, encoding local similarities among patients, and is visualized on the screen, producing a visual map of the patient distribution. Multidimensional glyphs are put in place of the nodes, revealing the properties that characterize each graph area. The graph construction method is extended to an incremental scheme, allowing federated graph formation. The proposed method is demonstrated in three use cases, regarding frailty in older adults, Sjögren's Syndrome patients, and a large-size diabetes dataset.
computer science, information systems, theory & methods,engineering, electrical & electronic, software engineering