In times of Big Data and Artificial Intelligence, decision-makers are faced with the problem of ever-increasing amounts of data and their correct interpretation. This development can be observed not only, but especially in the healthcare sector with the introduction of hospital information systems or the telematics infrastructure. Here, all actors in the German health care system, such as hospitals, pharmacies, and physicians in private practice, are integrated into a digital application. The Chair of Health Care Operations/Health Information Management tries to apply methods of Operations Research/Management Science especially, but not exclusively, in the health care sector to analyze the available data and to derive appropriate recommendations for action.
Data Analytics in Disaster Medicine
Due to the Corona pandemic, Bavaria was in a statewide disaster situation for three months in the spring of 2020. The declaration of a disaster situation is not uncommon, at least on a regional level, when one thinks of natural disasters, terrorist attacks, or mass traffic accidents. Disasters not only pose special challenges for the authorities but also for the healthcare system since a large number of injured or sick people are usually to be expected. For so-called multiple-casualty incident (MCI), a large number of precautionary algorithms have already been developed for prioritizing or triaging the urgency of treatment. Most of the algorithms have their origin in practice. The Chair of Health Care Operations/Health Information Management tries to take a data perspective and to evaluate existing algorithms, but also to develop new algorithms, for example with machine learning methods.
From life sciences to business management: simultaneous inferential statistics
If several null hypotheses are decided simultaneously based on a data set, there may be an inflation of the error of the first kind and thus of the false-positive results. Therefore, statisticians and later especially medical scientists have developed a variety of methods for such multiple testing problems in the last 80 years. Despite their high relevance, these methods find comparatively little application in empirical research as we can demonstrate in an analysis of high-level economic publications. In particular, the medical context of current research, the multiplicity of methods, and the two-step selection process seem to be obstacles. We address the first two issues by conducting extensive simulation studies to extract method recommendations for the managerial user. Regarding the two-step selection process, we develop two new multiple testing methods (SiMaFlex and SteMaFlex) that can transform the two-step decision process into a one-step decision. Performance analyses show that these new methods dominate existing methods. By focusing on these application-theoretic problems, we try to transfer the research area of simultaneous inference from medicine to economic research and thus make published results even more reliable.
Data Envelopment Analysis
Data Envelopment Analysis (DEA) is the most commonly used method for measuring the efficiency of non-profit organizations. It can be used, for example, to compare the performance of hospitals or universities. Best practice examples can be identified and learned from. Since the invention of the DEA method, a variety of models have evolved. However, which of these models provides the most accurate and robust results is still unclear. With our research, we try to answer these questions to enable reliable efficiency measurement in healthcare. Crucial for this benchmarking of different DEA models is the generation of artificial data that can be used to compare the model results. Monte Carlo simulation is used to generate a large number of different scenarios that differ, among other things, in the actual efficiency of the units, the number of inputs used, and the type of returns to scale at hand. After generating several test instances that ensure the robustness of the results, various performance indicators are used to determine the accuracy of the DEA models.
Away from the identification of the best models, a literature review on applications of the DEA method in health care was written in the area of Data Envelopment Analysis. In addition to numerous useful descriptive statistics, this also provides a guide to publications with important methodological findings.