HORIZON-HLTH-2027-02-TOOL-01-two-stage - Collaboration Offer

As Doğuş Teknoloji, we can develop data-driven and artificial intelligence–based solutions to support data integration, analysis, validation and regulatory readiness in biomarker-focused clinical research.

• Combine heterogeneous data from different sources into a common standard by establishing a data harmonisation pipeline and statistically report the effects of these variables on disease progression through exploratory data analyses.

• Combine different data types by using ensemble learning models (Random Forest, XGBoost) and generate clinical outcomes by optimising models on existing high-quality clinical data or real-world data (RWD).

• Make model decisions transparent by using explainable artificial intelligence libraries such as SHAP or LIME, measure model accuracy and reliability through performance metrics, and design simple visualisation interfaces for clinical assessment of biomarkers. Leverage LLM-based solutions to support model decision explanations.

• Validate models by conducting tests not only on training data but also on independent datasets from different centres and technically document clinical accuracy and positive impact on patients.

• Apply anonymisation techniques in data processing workflows in compliance with existing data privacy regulations and prepare the necessary technical performance documentation to support artificial intelligence tools in regulatory processes.