R in Industry: Biopharmaceutical Process Optimization through Advanced DoE

In our May 2025 meetup, we explore the interdisciplinary application of data science in the biopharmaceutical industry. Christina Yassouridis, Senior Data Scientist from Boehringer Ingelheim, will share with us how she designs and develops an R package which applies Design of Experiment (DoE) approaches to optimize biopharmaceutical manufacturing process.

Expect to hear about:
🟡 Data science: Theoretical background of advanced Design of Experiments (DoE) approach using Halton designs and Latin Hypercube Designs
🟣 R programming: The design and development of R package to meet industry needs
🟢 Biopharmaceutical industry: Current industry applications of R package in solving optimization problems

It will be an interesting sharing to see how R is applied to solve real-world problems in biopharmaceutical manufacturing, an area closely connected to our well-being. The theoretical background of the DoE approaches involved will also be introduced, which allows R users without field knowledge to follow along.

R-Ladies Vienna promotes gender diversity in the R community. All genders and skill levels in R are welcome to our events. We look forward to seeing you!

🗣️ Speaker
Christina is currently a Senior Data Scientist at Boehringer Ingelheim, one of the world’s largest pharmaceutical companies. With a PhD in Applied Statistics, Christina has extensive experience as a data science professional in both industry and consulting, and as a researcher at BOKU.

ℹ️ Abstract
Biopharmaceutical manufacturing encompasses complex processes involving living organisms and multiple interacting steps. To optimize these processes, data scientists develop digital representations using machine learning models to identify optimal process parameters across multidimensional experimental spaces. For Gaussian process regression models, evenly distributed experimental points yield superior results.

The R package {spacefillngDoEs}, developed by Christina, integrates and extends two established approaches:

1. Halton designs: These low-discrepancy sequences enable efficient quasi-Monte Carlo methods for numerical integration and sampling in high-dimensional spaces.
2. Latin Hypercube Designs: This statistical method generates near-random parameter value samples from multidimensional distributions.

By combining functionalities from the R packages {qrng} and {SLHD} while extending their capabilities to handle mixed data types, {spacefillngDoEs} addresses specific industry requirements.

In this presentation, Christina will introduce the theoretical foundations of generalized Halton designs, demonstrate the R package's functionality, and showcase its current industrial application.

⏱️ Duration
The event will last approximately 1 to 1.5 hours, including a Q&A session. Doors open at 17:45.

🗺️ Location
Seminar Room DA green 04 / Seminarraum DA grün 04 (DA04E10)
4th floor, green area
TU Wien Freihaus (map)
Wiedner Hauptstraße 8-10, 1040 Wien

📍 How to find us
In the green area of TU Wien Freihaus, take the lift to the 4th floor. After exiting the lift, turn left to enter through the green doors, then turn left again. The seminar room will be on your left. Use this floor plan to help locate the room.