Clinical Research and Real-World Evidence
At CARE, we want to improve patient outcomes by analyzing electronic health record data (EHR) in a multidisciplinary team of informaticians, statisticians and clinicians. We compile and utilize routinely collected patient data to answer clinical research questions. CARE is a cooperation between the Department of Hematology & Stem Cell Transplantation and the Institute for Artificial Intelligence in Medicine (IKIM). It is led by Dr. med. Christopher M. Sauer, M.D. M.P.H. Ph.D. and currently includes Mr. Gernot Pucher, M.Sc. M.Sc. as a PhD student as well as rotating MD students and research assistants.
CARE uses the Smart Hospital Information Platform (SHIP) to access all electronic health records from Essen University Hospital. Available in the standardized FHIR HL7 format, SHIP provides a rich and granular real-world data set – however it has not been comprehensively validated for clinical research. Thus, our first step is to clean, restructure and quality check the EHR data. After careful data validation, the target cohort is defined, relevant data is extracted, and statistical analyses are performed. We use a wide variety of statistical models to answer clinical questions, ranging from regression models and survival analysis to clustering and reinforcement learning. A special focus is on causal inference, whereby we disentangle data associations from true, targetable causal relationships. We strongly believe that our research outputs need to be beneficial to our patients and society. Therefore, we develop novel insights and algorithms that will be deployed at the patient’s bedside.
CARE supports clinicians and patients by providing them with the best available evidence and decision support tools. To this end CARE is fully committed to open science, meaning that we publish our code to encourage reproducibility by colleagues.
Besides hospital EHR data, the lab has a track record of analyzing intensive care data, primarily using publicly available datasets from the U.S. (MIMIC-IV, eICU-CRD) and Europe (AmsterdamUMCdb). For this, we collaborate with multiple research teams worldwide, e.g., at Harvard Medical School, Massachusetts Institute of Technology, Amsterdam UMC, Virginia Tech, Ghent University or Bergen University.
Leveraging Electronic Health Records for Data Science: Common pitfalls and how to avoid them. Sauer CM, Chen LC, Hyland S, Girbes A, Elbers P, Celi LA. Lancet Digit Health. 2022 Dec;4(12):e893-e898.
Systematic review and comparison of publicly available ICU data sets – A decision guide for clinicians and data scientists. Sauer CM, Dam TA, Celi LA, Faltys M, de la Hoz MAA, Adhikari L, Ziesemer KA, Girbes A, Thoral PJ, Elbers P. Crit Care Med. 2022 Mar 2. - Volume - Issue -DOI: 10.1097/CCM.0000000000005517
Understanding critically ill sepsis patients with normal serum lactate levels – results from U.S. and European ICU cohorts. Sauer CM, Gomez J, Botella MR, Ziehr DR, Oldham WM, Gavidia G, Rodriguez A, Elbers P, Girbes A, Bodi M, Celi LA. Nature Sci Rep. 2021 Nov:20076. DOI: 10.1038/s41598-021-99581-6.
Machine learning can accurately predict pre-admission baseline hemoglobin and creatinine in intensive care patients. Dauvin A, Donado C, Bachtiger P, Huang KC, Sauer CM, Ramazzotti D, Bonvini M, Celi LA, Douglas MJ. NPJ Digit Med. 2019 Nov 29;2:116. DOI: 10.1038/s41746-019-0192-z.
Brehmer, A., Sauer, C.M., Salazar, J., Herrmann, Kelsey, Kim, M., Keyl, J., Bahnsen, F.H., Frank, B., Köhrmann, M., Rassaf, T., Mahabadi, A.-A., Hadaschik, B., Darr, C., Herrmann, Ken, Tan, S., Buer, J., Brenner, T., Reinhardt, H.C., Nensa, F., Gertz, M., Egger, J., Kleesiek, J., SSRN 2023. Establishing Medical Intelligence - Leveraging FHIR to Improve Clinical Management. https://doi.org/10.2139/ssrn.4493924