Learning Circadian Treatment Patterns from Real-World Data: A Medical Informatics Approach to Chronotherapy

• Applicant: Dr. Susanne Ibing
• Subject Area: Medical Informatics and Medical Bioinformatics ; Gastroenterology
• Term: since 2026
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 576382343

Electronic health records (EHRs) hold immense, yet underutilized potential for generating new, data-driven insights to improve therapeutic decision-making. In particular, the use of time-resolved clinical routine data, such as the precise timing of medication administration, offers new opportunities to detect and analyze dynamic treatment effects in real-world care settings. However, the field of medical informatics currently lacks a generalizable methodological framework for identifying and modeling such time-dependent treatment patterns from real-world data. This project aims to develop a scalable framework for modeling circadian effects in clinical routine data. It will leverage and advance state-of-the-art methods in medical informatics, including Bayesian modeling, time series analysis, machine learning, and natural language processing (NLP). A particular focus is the integration of individualizable circadian features (e.g., chronotype, activity patterns), initially from wearable data and subsequently from EHRs, as well as social determinants of health (SDoH) extracted from clinical free-text data. The developed methodology will be applied to a highly relevant medical question: the potential impact of infusion timing on the treatment response to biologics in inflammatory bowel disease (IBD). The project builds on a pilot study involving over 1,100 IBD patients from the Mount Sinai Health System, in which a significant association was found between later infusion times and increased non-response rates. Building on these results, we plan to extend the modeling approach and integrate additional endpoints and covariates in a multi-center study design using datasets from the U.S. and Europe. The project pursues two main objectives: 1. Method Development: Estimating individual circadian rhythms from EHRs. We will develop the underlying models using EHR-linked wearable data. 2. Chronotherapeutic Effects: Analyzing circadian influences on biologic treatment response while accounting for SDoH and patient chronotypes. This research combines technological innovation with clinical relevance and creates a methodological foundation for integrating circadian medicine into data-driven healthcare strategies. It contributes to the development of personalized, time-optimized therapies supported by real-world evidence.

• DFG Programme: WBP Fellowship
• International Connection: USA
• Host: Francesca Petralia, Ph.D.