KNAVE: Knowledge-based Navigation of Abstractions for Visualization and Explanation

Introduction: KNAVE II is a tool for the interpretation, summarization, visualization, explanation, and interactive navigation in a context-sensitive manner through time-oriented raw clinical data and higher-level concepts abstracted from the time-oriented data.

A view of an individual patient's data in the KNAVE prototype. On the left hand side, a browser to the clinical domain's ontology, coming from the knowledge base, is shown. The user selects a raw data or abstract concept, which is then retrieved or computed on the fly and displayed as a panel on the right hand side. Operators represented as icons, in each panel, enable the user to perform actions such as semantically explore the concept (e.g., through IS-A and ABSTRACTED-FROM relations, as shown in the exploration window in the right hand bottom corner), query the knowledge used to derive the concept, and add statistics about raw and derived concepts (see 1st and 4th panels from the top). Users can also zoom into and out of all panels simultaneously or of one panel by changing the temporal granularity, using the timeline under each panel or the global time-unit scale bar.

KNAVE II is an application that enables visual browsing and exploring of patient's data. Not only does it enable browsing the raw patient data, but it also enables viewing its intelligent interpretation, as will be explained shortly.
Data consists of both raw measured values (i.e., lab tests results) and external interventions (i.e. medications and procedures).

In order to interpret raw data the predefined knowledge base is used. The knowledge base defines three types of interpretations, as follows:

1. Classification of data (i.e., low – normal– high hemoglobin
   state)
2. Change of data (i.e., rising – decreasing glucose levels)
3. Rate of data change (i.e., slow, fast)

In addition, there are interpretations that depend on other interpretations, such as bone marrow toxicity, which is defined by the RBC’s and platelets count level.
The knowledge base definitions are context-sensitive (i.e., age and gender).

KNAVE Team personal: David Boaz, Gil Tahan, Maya Galperin

Related projects: The Converter, KAT, Mediator

Application projects exemple:

Back to Research Projects main page