KNAVE-II: A distributed architecture for interactive visualization and intelligent exploration of time-oriented clinical data

Interpretation and exploration of longitudinal clinical data is a major part of diagnosis, therapy, quality assessment, and clinical research, particularly for chronic patients. KNAVE-II is an intelligent interface to a distributed architecture specific to the tasks of query, knowledge-based interpretation, summarization, visualization, interactive exploration of large numbers of distributed time-oriented clinical data and dynamic sensitivity analysis of these data. The web-based architecture enables users (e.g., physicians) to query, visualize and explore clinical time-oriented databases. Both, the generation of context-sensitive interpretations (abstractions) of the time-stamped data, as well as the dynamic visual exploration of the raw data and the multiple levels of concepts abstracted from these data, are supported by runtime access to domain-specific
knowledge bases, maintained by domain experts. KNAVEII was designed according to a set of well-defined desiderata. The architecture enables exploration along both absolute (calendar-based) and relative (clinically
meaningful) time-lines. The underlying architecture uses standardized vocabularies (such as a controlled dictionary for laboratory tests and physical observations), and predefined mappings to local data sources, for communication among its various components. Thus, the new framework enables users to access and explore multiple remote heterogeneous databases, without explicitly knowing their local structure and vocabulary, through a filter of a set of task-specific knowledgebases. The complete architecture has been implemented and is currently evaluated by expert clinicians in several
medical domains, such as oncology, involving monitoring of chronic patients.