Aus Aifbportal
Wechseln zu:Navigation, Suche

Automating Discharge Summaries Based on the MIMIC III Data Set

Information on the Thesis

Type of Final Thesis: Bachelor, Master
Supervisor: Richard Guse
Research Group: Critical Information Infrastructures

Archive Number: 5.093
Status of Thesis: Open
Date of start: 2023-09-15

Further Information

Sorry, no english description available!


Discharge summaries play an important role in hospitals for patients’ aftercare. It includes important diagnoses that are relevant for further ambulatory care and attending general practitioners. Furthermore, it contains prescriptions for medicine and medical devices that patients should follow after discharge. However, despite its importance for the further treatment and recovery process for patients, discharge summaries are often erroneous. This eventually leads to further delay and lower care quality received for a patient after leaving the hospital. In severe cases, it might even lead to a worsening of a patient’s condition. A main reason for erroneous discharge summaries is that physicians and nurses in hospitals often do not have enough time for documentation. When patient care is the highest priority, additional documentation is often perceived as a burden. To overcome this issue, clinical support systems must be developed to both automate the creation of discharge summaries and ensure the quality of discharge summaries.


- Create an overview of what data of patients’ stay documentation is contained in their respective discharge summary

- Design and develop a tool that creates a first discharge summary draft

- Design and develop a tool that supports a physician in the creation of discharge summaries based on the patient data available

Note: This is an umbrella topic. The overall goal, context, and direction of the thesis are defined in the first kickoff meeting.

Research Method:

Statistical Analysis based on Natural Language Processing (NLP), Design Science Research Methodology or similar.

Note: Other techniques such as transfer learning or Large Language Models (LLMs) can also be used.


Callen, J., McIntosh, J., & Li, J. (2010). Accuracy of medication documentation in hospital discharge summaries: A retrospective analysis of medication transcription errors in manual and electronic discharge summaries. International Journal of Medical Informatics, 79(1), 58–64.

Johnson, A. E. W., Pollard, T. J., Shen, L., Lehman, L. H., Feng, M., Ghassemi, M., Moody, B., Szolovits, P., Anthony Celi, L., & Mark, R. G. (2016). MIMIC-III, a freely accessible critical care database. Scientific Data, 3(1), Art. 1.

Moldskred, P. S., Snibsøer, A. K., & Espehaug, B. (2021). Improving the quality of nursing documentation at a residential care home: A clinical audit. BMC Nursing, 20(1), 103.

Starlinger, J., Kittner, M., Blankenstein, O., & Leser, U. (2017). How to improve information extraction from German medical records. Information Technology, 59(4), 171–179.