A health data lake that became the foundation for AI drug discovery, antibiotic resistance research, and a Microsoft flagship case study

Healthcare generates vast quantities of structured and unstructured data - lab results, EKG readings, free-text clinical notes, wearable device outputs - held across siloed systems. This data is essential for clinical operations but practically inaccessible for research at scale. The University of Pécs needed infrastructure to collect, standardise, and make this data jointly queryable - enabling high-impact research and AI model development without manual data preparation for every study.

E-Group built the InnoHealth DataLake with the University of Pécs starting in 2017 (formal launch 2018), using its Smart Data / DLX platform on a Microsoft SQL Server Big Data Cluster. The four-layer architecture (Data, Analytical, Knowledge & Research, and Unified Management layers) aggregates clinical data from structured and unstructured sources - free-text medical records, lab results, EKG monitors, wearable devices - applying automated ETL to standardise to the OMOP Common Data Model. A multimodal monitoring sub-project produced a predictive algorithm for migraine and epilepsy patients with an error rate below one event per hour. Principal Architect: Alexandra Babaics, who presented the project at Microsoft's SQL Server 2019 Uncovered Virtual Event.

The InnoHealth DataLake became the foundation for multiple downstream projects: Repo-X (AI drug repositioning, with a prostate cancer candidate in Phase 2 clinical trials), the Animal Health Innovation Platform (antibiotic resistance), and multimodal patient monitoring research. Microsoft selected the implementation as a flagship SQL Server 2019 case study.