Skip to main content
Aurélien Vannieuwenhuyze

Demo 001 · Available

Making hidden temporal relations explicit in a clinical trajectory

This interactive demonstrator turns part of the work presented in A Conceptual Model for Discovering Implicit Temporal Knowledge in Clinical Data into an interface. It shows how timestamped clinical events can become an explicit, queryable, and explainable temporal representation.

The scenario

A synthetic patient case around a bloodstream infection

The demo presents a fully synthetic ICU trajectory around T0, a positive blood culture. It includes biomarkers, clinical signs, a microbiology event, and antibiotic therapy.

The scenario reflects several competency questions from the paper: which biomarkers increase before a positive blood culture? Which early signs precede confirmed infection? How long is the delay between a fever spike and the first antibiotic?

Interactive journey

Four steps from data to temporal knowledge

01

Patient record

Explore the clinical trajectory around T0 and inspect event values and timestamps.

02

Structure time

Distinguish point events from durative events and anchor them as time:Instant or time:Interval.

03

Explore the graph

Visualize computed relations between events and read their clinical-language translation.

04

Reveal knowledge

Surface trends, delays, and sequences that were not directly stored as such in the source data.

What the prototype demonstrates

Time becomes an explicit structure, not just a timestamp

The method links clinical events to OWL-Time temporal entities, then computes temporal relations and distances. The graph can therefore be enriched with facts such as “before”, “within the previous 48 hours”, or an exact delay in hours.

In the demonstrator, this logic makes it possible to visualize biomarker progression, a delay between fever and antibiotic therapy, and an ordered clinical sequence around the positive blood culture.

Research connection

From the scientific paper to an interactive demonstrator

The published work proposes a competency-question-driven methodology for transforming intensive-care data into a temporal RDF graph: concept definition, event extraction, OWL-Time annotation, graph construction, and temporal-relation enrichment.

The demonstrator does not expose the MIMIC-IV data used in the study. It deliberately uses a synthetic patient so that the method can be illustrated publicly and interactively.

Keep in mind

A research demonstration, not a medical device

The displayed data and conclusions are synthetic. The prototype illustrates a temporal representation and reasoning method; it does not diagnose infection, recommend treatment, or support clinical decisions.

Try the prototype

Explore the trajectory and temporal graph yourself

The demonstrator is interactive and publicly available. No account is required.