This thesis explores a scalable alternative by asking whether publicly available hospital documents can be mined to derive structured, framework-aligned signals of digital maturity for Swiss health institutions. Using a Design Science Research (DSR) methodology, the study develops an end-to-end NLP artifact that ingests heterogeneous sources, cleans and segments them into analysis-ready passages, and embeds the resulting text with a multilingual sentence-transformer to enable languageagnostic semantic retrieval. Building on this retrieval layer, a Large Language Model (LLM) is used to classify evidence-bearing text segments into the seven dimensions of the DigitalRadar Krankenhaus (DRK) framework through prompting and structured outputs supported by extractive evidence spans, yielding transparent and auditable maturity signals.

The artifact is evaluated through both intrinsic and extrinsic validation. Intrinsically, a stratified manual review of classified segments indicates that the pipeline produces largely plausible DRK assignments. Extrinsically, hospital-level DRK proxies arecorrelated with operational indicators from the Swiss Federal Statistical Office under literature-motivated hypotheses.

Overall, the thesis demonstrates that automated, document-based maturity assessment is feasible and interpretable as an evidence-driven proxy, and it motivates a hybrid evaluation strategy that combines scalable text analytics with selective validation using direct capability metrics and expert review.