Health and Environmental Analytics
Analyzing data to derive interpretable results using statistical and machine learning techniques
We use a diverse mix of statistical and machine learning approaches as well as methods for data-based model development to analyze complex data sets and predict trends and future outcomes. The ultimate goal is to transform raw data into actionable insights that guide domain and policy decisions.
Would you like to discuss specific aspects of these methods or learn more details about how they can be applied in real-world scenarios?
Our Data Analytical Approaches
- Statistical Methods are techniques used to collect, analyze, interpret, and present data in order to uncover patterns, test hypotheses, and make informed decisions under uncertainty.
- Predictive Analytics involves using statistical models and machine learning techniques, including ensemble and deep learning, to analyze current and historical data in order to make predictions about future outcomes or trends.
- Special Topics
- Causal Inference serves for identifying and quantifying cause-and-effect relationships from data, beyond mere associations or correlations.
- Conformal Prediction is a way of constructing reliable measures of uncertainty for each prediction.
- Data Anonymization protects sensitive personal data, ensures compliance with privacy regulations, reduces risks, and enables secure, innovative data sharing and analysis.
- Experimental and Study Design for planning and structuring of research to systematically investigate hypotheses, ensuring valid and reliable results through controlled methods and procedures.
- Network Analysis is a method for examining the structure and dynamics of complex systems by representing entities as nodes and their relationships as edges in a graph.
Research & Projects
A Model-Based Three-Stage Classifier for Particulate Matter
To enable Scanning Electron Microscopy with Energy Dispersive Spectroscopy (SEM-EDS) for routine analysis, automated and standardized evaluation of large particle datasets is essential. In collaboration with Particle Vision GmbH, we have developed a universally applicable three-stage particle classifier that categorizes particles into thousands of classes based on their chemical composition.
Deep Learning-Based Classification of Histological Subtypes of Lung T
b) VGG CNN extracts 4,096 image features at specified layer.
More in project details.
Subtyping lung tumors is critical for treatment selection and is currently performed by pathologists through visual inspection of Hematoxylin and Eosin stained sections. In this study, we analyze 50 histological images per patient from 207 diagnosed cases to distinguish adenocarcinoma and squamous cell carcinoma. Deep learning methods, specifically convolutional neural networks, enable automated classification with accuracy comparable to that of expert pathologists.
Tariff System for Inpatient Rehabilitation
As part of the performance-based and lump-sum reimbursement model for hospitals and clinics mandated by the Swiss Health Insurance Act, ZHAW has developed a tariff system for performance-based daily flat rates in rehabilitation.
Data Anonymization
For both data sharing and internal use, anonymization is required to prevent the identification of individuals while preserving the analytical quality of the data as much as possible. The data anonymization process is applied to complex datasets provided by Helsana.
Prognolite – Sales Forecasting for the Restaurant and Catering Industry
Prognolite is a startup that provides software for forecasting sales and customer frequency in the restaurant and catering industry. We developed prediction algorithms that work with inputs such as calendar data, weather, events, and more—enabling efficient decision-making and helping to reduce food waste.
Publications
Articles in scientific journals, peer-reviewed
Spurk Christoph, Koch Carmen, Bürgin Reto, Chikopela Louis, Konaté Famagan, Nyabuga George, Sarpong Daniel Bruce, Sousa Fernando, Fliessbach Andreas, 2023. Farmers’ innovativeness and positive affirmation as main drivers of adoption of soil fertility management practices: evidence across sites in Africa. The Journal of Agricultural Education and Extension. https://doi.org/10.1080/1389224X.2023.2281909
Bürgin Reto, Muratori Corrado, Roca-Riu Mireia, Heitz Christoph, 2023. A space-time model for demand in free-floating carsharing systems. Journal of Advanced Transportation. 2023(6610624). https://doi.org/10.1155/2023/6610624
Arpogaus Marcel, Voss Marcus, Sick Beate, Nigge-Uricher Mark, Dürr Oliver, 2023. Short-term density forecasting of low-voltage load using bernstein-polynomial normalizing flows. IEEE Transactions on Smart Grid. 14(6), S. 4902-4911. https://doi.org/10.1109/TSG.2023.3254890
Mildenberger Thoralf, Braschler Martin, Ruckstuhl Andreas, Vorburger Robert, Stockinger Kurt, 2023. The role of data scientists in modern enterprises : experience from data science education. SIGMOD Record. 52(2), S. 48-52. https://doi.org/10.21256/zhaw-27357
Müller Werder Claude, Mildenberger Thoralf, Steingruber Daniel, 2023. Learning effectiveness of a flexible learning study programme in a blended learning design : why are some courses more effective than others? International Journal of Educational Technology in Higher Education. 20(10). https://doi.org/10.1186/s41239-022-00379-x
Articles in scientific journals, non-peer-reviewed
Thalmann Basilius, Hofer Christoph, Wächter Daniel, Kulli Beatrice, 2022. Per- und polyfluorierte Alkylsubstanzen (PFAS) in Schweizer Böden. altlasten spektrum. 31(6), S. 176-179. https://doi.org/10.37307/j.1864-8371.2022.06.05
Published Project Reports
Bürgin Reto, Stucki Michael, Vetsch-Tzogiou Christina, Kauer Lukas Kohler Andreas, Drewek Anna, Thommen Christoph, Dettling Marcel, Wieser Simon, 2024. Wirkungsanalyse zum Risikoausgleich mit pharmazeutischen Kostengruppen (PCG): Schlussbericht. https://doi.org/10.21256/zhaw-30489
Drewek Anna, Ordelt Christian, Riahi Nima, Sedding Helmut, 2024. 100 Jahre Sollzeiten - Ein Konzept für die Zukunft?. Logistics Innovation. 2024(1), S. 10-13.
Cieliebak Mark, Drewek, Anna, Jakob Grob Karin, Kruse Otto, Mlynchyk Katsiaryna, Rapp Christian, Waller Gregor, 2023. Generative KI beim Verfassen von Bachelorarbeiten: Ergebnisse einer Studierendenbefragung im Juli 2023. https://doi.org/10.21256/zhaw-2491
Talks, peer-reviewed
Bürgin, Reto; Vetsch-Tzogiou, Christina; Stucki, Michael; Kauer, Lukas; Pirktl, Lennart; van Kleef, Richard C.; Kohler, Andreas; Drewek, Anna; Thommen, Christoph; Dettling, Marcel; Wieser, Simon, 2024. Improving risk adjustment in Switzerland with pharmaceutical cost groups [Paper]. In: 6th Swiss Health Economic Workshop, Lucerne, Switzerland, 7 June 2024.
Teaching
“We teach students to learn from data with statistical methods, separating structure from noise."
Lectures
We teach Bachelor students all about statistical data analysis starting from exploratory data analysis, to inferential statistics, statistical modelling, advanced regression modelling, survey design analysis, data mining, predictive modelling, data analytics, and statistical quality control.
We are teaching mainly in the Bachelor programs Wirtschaftsingenieurwesen, Mobility Science and Data Science and two module courses Advanced Statistical Data Analysis and Business Analytics in the Master of Engineering program.
Furthermore, we are running two CAS course call Data Analysis and Advanced Statistical Data Analysis organised by the school of Engineering under the program MAS Data Science.
Supervising Student Projects
We are excited to collaborate with students and provide dedicated support to both Bachelor's and Master's (MSE) candidates throughout their project journeys.
You can explore current project and thesis opportunities on Complesis.
We also welcome your own ideas and would be glad to discuss them with you!
Team
-
ZHAW School of Engineering
Technikumstrasse 81
8400 Winterthur
