Medexter Presents Three Scientific Papers at MIE 2026

Medexter presented three scientific papers at MIE 2026 in Genoa, demonstrating how semantic technologies, clinical knowledge, and artificial intelligence can be combined to solve real-world healthcare challenges. The work ranged from automated microbiology data interpretation and ontology management to the integration of agentic LLMs with ArdenSuite-based clinical decision support.

Automated Differential Time to Positivity Analysis

Our first paper focused on automating Differential Time to Positivity (DTP) calculations for the diagnosis of catheter-related bloodstream infections (CRBSIs).

Using data from our Momo platform, we combined semantic NLP, ontology-driven normalization, and fuzzy matching techniques to automate the analysis of microbiology results. The approach supports clinicians in distinguishing CRBSIs from secondary bacteremia and demonstrates how semantic technologies can transform routine laboratory data into actionable clinical knowledge. Find the paper here.

Auto Ontology: Towards Automated Term-to-Concept Assignment in Microbiology Analytics

Maintaining clinical microbiology ontologies requires significant effort, particularly when heterogeneous laboratory terminology must be mapped to standardized concepts.

To address this challenge, we presented Auto Ontology, a lightweight embedding-based approach that generates concept suggestions for ontology curators. The method supports semi-automated ontology maintenance while remaining suitable for deployment on standard hospital infrastructure without GPUs or external cloud services. Find the paper here.

Integrating Agentic LLMs with Clinical Decision Support

Our third contribution explored how lightweight agentic large language models can be combined with established clinical decision support systems.

We connected agentic LLMs to our ArdenSuite software through the Model Context Protocol (MCP) and evaluated the approach using hepatitis serology test result interpretation. By combining semantic AI capabilities with deterministic clinical knowledge, the solution provides structured and transparent interpretations while supporting reliable use in real-world clinical environments. Find the paper here.