The Chair of Health Care Operations/Health Information Management is pursuing several projects related to the Covid-19 pandemic, analysing various issues in collaboration with high-level practice partners using a variety of methodological approaches.

Development of a forecasting tool on capacity utilization by COVID-19 patients in hospitals

The COVID-19 pandemic is characterized by a slow-building demand on healthcare resources with local hotspots, posing enormous problems for the healthcare system. For hospitals, one of the biggest challenges is maintaining bed capacity, especially since bed demand is difficult to predict throughout a pandemic. To assist decision-makers, a simulation-based capacity utilization forecasting tool has been developed with the University Hospital Augsburg to estimate the required bed capacities under different pandemic courses.

 

Current knowledge about the course of the spread, in particular the growth rate of cumulative new infections per day, is used as input. To reflect uncertainty, distribution functions based on real data of the growth rate of cumulative new infections, the length of stay, and the proportion of COVID-19 patients requiring hospitalization in the region are modelled. This is followed by a Monte Carlo simulation, providing an estimate of the required bed capacity for several days in the future.

 

With the help of the simulation-based forecast of the capacity utilization, clinics and emergency management command centers can be provided with valuable information for estimating the short-term development of capacity requirements for suspected cases as well as confirmed COVID-19 patients. The practical application of the method at the University Hospital Augsburg shows reliable results. If the political guidelines regarding contact and curfew periods were to be changed in the future, the course of the required bed capacities could be forecast more accurately, based on the growth rate of cumulative new infections, which would then be within the historically recorded range.

 

We are currently using the forecasting tool to create reports for the Bavarian Ministry of Health and the Rescue Association of Swabia to support their political and operational work.


If you are interested in our tool, please feel free to contact   Dr. Jan Schoenfelder.

 

Cooperation:

  • Lehrstuhl Brunner (Wirtschaftswissenschaftliche Fakultät)
  • Bayerisches Staatsministerium für Gesundheit und Pflege
  • Prof. Dr. Heller (Universitätsklinikum Augsburg)

 

Publications:

Römmele, C., Neidel, T., Heins, J., Heider, S., Otten, V., Ebigbo, A., Weber, T., Müller, M., Spring, O., Braun, G., Wittmann, M., Schoenfelder, J., Heller, A. R., Messmann, H., & Brunner, J. O. (2020). Bettenkapazitätssteuerung in Zeiten der COVID-19-Pandemie: Eine simulationsbasierte Prognose der Normal- und Intensivstationsbetten anhand der deskriptiven Daten des Universitätsklinikums Augsburg. Der Anaesthesist, 1–8. Advance online publication.  https://doi.org/10.1007/s00101-020-00830-6

AI-based prediction of COVID-19 using laboratory results

The pandemic caused by the new coronavirus (SARS-CoV-2) poses enormous problems for the health care system as a whole and hospital capacities in particular. For hospitals and emergency departments, one of the greatest challenges is managing patient flow, as a differential diagnosis of possible COVID-19 infection of symptomatic patients is difficult due to the broad clinical presentation of the COVID-19 disease.  Providing timely and objective guidance for deciding whether a COVID-19 disease is present could avoid unnecessarily burdening the COVID bed capacity with COVID-19 negative patients.

 

Based on the laboratory values of the previously confirmed and also ruled out COVID-19 patients collected at the University Hospital Augsburg, classical machine learning algorithms as well as a newly developed algorithm, which we call COVIDAL, were trained. We are already applying the algorithms in an Excel-based solution as well as a browser-based tool, the COVIDAL APP.  Since analyses show that a broader data set may significantly improve the algorithms' sensitivity and specificity, we are now pursuing a multicenter approach. By integrating relevant data sets from different hospitals in Germany as well as from the LEOSS registry, we expect to further improve the results. These will be validated in a head-to-head comparison with several intensive care physicians or infectious disease specialists familiar with the treatment of COVID-19.

 

The project will improve the management of patient flows in the hospital and provide physicians with a real-time decision support tool for the differential diagnosis of a possible COVID-19 infection.

For more information about our algorithms and the application tool, please feel free to contact  Dr. Christina Bartenschlager.

 

Cooperation:

  • Lehrstuhl Brunner (Wirtschaftswissenschaftliche Fakultät)
  • Prof. Dr. Hoffmann (Universitätsklinikum Augsburg)
  • Prof. Dr. Heller (Universitätsklinikum Augsburg)
  • Prof. Dr. Messmann (Universitätsklinikum Augsburg)

 

Working Paper:

Bartenschlager CC, Ebel SS, Kling S, Brunner JO, Heller AR, Hoffmann R, Messmann H, Römmele C (2020): A sensitive combined machine learning algorithm for the prediction of Covid-19 infections in European hospitals. Working paper, University of Augsburg.

Covid-19 triage of symptomatic patients in the emergency department

With the second coronavirus wave in Germany, the discussion about a possible triage has come back into the focus of public interest. Although there are already some suggestions in the literature on how symptomatic patients should be triaged in the emergency department concerning treatment urgency, only a few consider the problem from a data perspective. Using data from the LEOSS registry, we validate existing concepts, try to identify corresponding optimization potentials, and to implement them. Furthermore, we use Monte Carlo simulations to investigate issues related to the triaging of corona patients.

If you are interested in the results of our study, please feel free to contact Dr. Christina Bartenschlager .

 

Cooperation:

  • Lehrstuhl Brunner (Wirtschaftswissenschaftliche Fakultät)
  • Prof. Dr. Messmann (Universitätsklinikum Augsburg)
  • Prof. Dr. Heller (Universitätsklinikum Augsburg)

To the performance of infection-prevention strategies in hospitals during Covid-19

FFP2 mask or surgical mask? Rapid test or PCR test? In this project, we are trying to answer such and similar questions with the help of statistical analyses, Monte Carlo simulations, and decision-theoretical approaches in a hospital environment, e.g. in an endoscopic unit. Various integrated key figures from a medical perspective, such as sensitivity or number of additional infections, and from a business perspective, such as cost aspects, are used for evaluation.

 

If you are interested in the results of our study, please feel free to contact Dr. Christina Bartenschlager.

 

Cooperation:

  • Lehrstuhl Brunner (Wirtschaftswissenschaftliche Fakultät)
  • Prof. Dr. Messmann (Universitätsklinikum Augsburg)

 

Publications and Working Paper:

Ebigbo A, Römmele C, Bartenschlager CC, Temizel S, Kling E, Brunner JO, Messmann H (2020): Cost-effectiveness analysis of SARS-CoV-2 infection-prevention strategies including pre-endoscopic virus testing and use of high-risk personal protective equipment. Endoscopy.

Kahn M, Schuierer L, Bartenschlager CC, Zellmer S, Frey R, Freitag M, Dhillon C, Heier M, Ebigbo A, Denzel C, Temizel S, Messmann H, Wehler M, Hoffmann R, Kling E, Römmele C (2020): Performance of antigen testing for diagnosis of COVID-19 – a direct comparison of a lateral flow device to nucleic acid amplification based tests. Working paper, University of Augsburg.

Visitor management in hospitals during times of a pandemic

The restrictive visiting regulations, which have been prescribed in the health sector since the beginning of the Corona pandemic, present hospitals with enormous challenges. To control possible entries of infections by visitors and externals, best-practice concepts are being developed for the University Hospital Augsburg. This is done within the framework of the Federal Research Network Applied Surveillance and Testing ( B-FAST).

 

Augsburg University Hospital introduced PLANFOX visit management from XITASO at a very early stage. The tool reduces the number of staff required for entry checks, enables rapid contact tracing in the event of infection, and data documentation in compliance with data protection regulations with extensive evaluation options. The goals of the B-FAST subproject Visitor Management, which is led by us and financed by federal funds, are the scientific evaluations of such admission concepts and corresponding prospective recommendations. The project will also address questions of dynamic adaptation of visitor admission to local and regional outbreak events while maintaining dignified palliative care, as well as the coordination of visitor flows and hygiene training.

 

To achieve this, planned methods include literature searches, nonrandomized controlled trials, semistructured interviews, structured questionnaires, qualitative and quantitative analyses, descriptive statistics, forecasting methods, Monte Carlo simulation, queuing theory, and simulation methods.

 

If you are interested in the results of our study, please feel free to contact  Dr. Christina Bartenschlager.

 

Cooperation:

  • Lehrstuhl Brunner (Wirtschaftswissenschaftliche Fakultät)
  • Prof. Dr. Messmann (Universitätsklinikum Augsburg)
© Netzwerk Universitäts Medizin

Search