From Manual Search to Artificial Intelligence: A Methodological Framework for Developing Patient Journey Models from Electronic Health Records Using Natural Language Processing

Abstract Improving care pathways for chronic conditions requires a holistic, data-driven understanding of the patient journey. This study introduces a novel framework that reconstructs and simulates patient journeys using structured medication data and probabilistic modeling. Chronic spontaneous urticaria (CSU) was selected as a use case, with symptom onset estimated via natural language processing (NLP) and treatment cycles identified through Anatomical Therapeutic Chemical (ATC) code sequences—an approach not previously applied in this context. A Markov model was built to simulate transitions across health states using de-identified EHR data from 74,664 CSU patients. Results revealed significant delays in diagnosis and variability in treatment initiation, with 29.7% of patients receiving no therapy. The model enables simulation of targeted interventions to streamline care, reduce delays, and optimize resource use. While intentionally simple, the framework is scalable across chronic conditions and supports value-based care through data-driven decision-making.