Cardiff University

Abstracts 

Understanding Healthcare Systems: Making Sense of Complexity with Data Science and Analytics

(Paul Harper)

Healthcare systems are stochastic in nature; that is they typically operate in an environment of uncertainty and variability, both at scale and within highly complex and connected networks. Furthermore, many healthcare services are under significant pressure to deliver more with less, whilst at the same time trying to recover from the challenges of the COVID-19 pandemic. Data Science and Analytics/Operational Research (OR) methods can help healthcare providers and policy makers move towards optimally configured services. In this talk I will briefly outline an innovative researchers-in-residence programme established in partnership between Cardiff University and NHS Wales. Using a range of data science and modelling techniques, both locally and more widely to support national policy, our research has directly led to evidenced cost savings whilst significantly improving patient outcomes and ultimately saving lives, in areas such as emergency care services, mental health and cancer care.

Bio:

Paul Harper is a Professor of Operational Research in the School of Mathematics at Cardiff University and Director of the Wales Data Nation Accelerator (WDNA). His research interests are in stochastic modelling, including queueing theory, simulation methods, optimisation and game theory, and applications to healthcare. Author of more than 100 peer-reviewed papers and book chapters, Paul has been a named investigator on over £14million of funded research grants, recipient of a Times Higher Education award for ‘Outstanding Contribution to Innovation and Technology’ and the UK OR Society’s Lyn Thomas Impact Medal. He is a founding editor-in-chief of the international journal Health Systems (Taylor & Francis), an elected Fellow of the Learned Society of Wales (FLSW), recipient of the 2018 "Companion of OR" Award (The OR Society) and was a panel member of the UK Government’s Research Excellence Framework (REF2021) for Mathematical Sciences.  

https://www.cardiff.ac.uk/people/view/98650-harper-paul

Application of Data Science in High-Hazard Industry: Learnings and Unique Challenges 

(Nyala Noe)

Process safety focuses on preventing the uncontrolled release of hazardous materials in order to avoid explosions, fires, or hazardous materials to be released into the environment. Examples of process safety accidents include the Deep Water Horizon oil spill of 2010, the Challenger explosion of 1986, or the nuclear disaster of Chernobyl, also in 1986. The high-hazard industry has focused on learning from past accidents. Process safety related incidents have been going down since the 80s through improvements to processes, training, and safety culture. However, they have been plateauing in the last decade. 

Bio:

Nyala Noë is the Senior Data Scientist at Empirisys. Empirisys is a startup that brings together data scientists and engineers to offer data science solutions to improve process safety in high-hazard industry (e.g., oil&gas, chemical refineries, nuclear). She has a background in social psychology and computer science, and has developed a particular interest in using networks to make sense of complex data. After completing her PhD at Cardiff University, she started her career as a data scientist 5 years ago and has been with Empirisys since its launch in 2020 as a founding employee. More details at: https://www.empirisys.io/ 

University of Bremen

Abstracts 

On Data and Learning in Optimization under Uncertainty 

(Nicole Megow)

Optimization under uncertainty is a key challenge in real-world decision-making. Often irrevocable decisions must be made for an initially unknown input that is revealed incrementally. In the advent of data-driven applications and given the success of machine-learning methods, the assumption of not having any prior knowledge about future input seems overly pessimistic. However, simply trusting them might lead to very poor solutions as these predictions come with no quality guarantee. In this talk we present recent developments in the young line of research that integrates such error-prone predictions into algorithm design to break through worst case barriers. We discuss algorithmic challenges with a focus on online routing and network design and present algorithms with provable performance guarantees depending on certain error metrics. 

Bio:

Nicole Megow studied mathematics and business administration at TU Berlin and the Massachusetts Institute of Technology, USA. She received her PhD in Mathematics from TU Berlin in 2006. She was postdoc and senior researcher at the Max Planck Institute for Informatics, Saarbrücken, held a position as interim professor for discrete optimization at TU Darmstadt 2011/12, and headed an Emmy Noether Research Group at TU Berlin starting 2012. Subsequently, she was an assistant professor for Discrete Mathematics at TU Munich. Since 2016 she holds the chair for Combinatorial Optimization in the Faculty of Mathematics and Computer Science at the University of Bremen.

Nicole Megow’s research interests lie at the intersection of discrete mathematics, theoretical computer science, and operations research. In particular, she works on the design and analysis of efficient algorithms for combinatorial optimization problems. Her research is focused on understanding how to cope with incomplete or imperfect information (uncertainty of problem data, complete lack of knowledge, untrusted predictions) when solving such problems. She contributes with theoretic results and applies them to complex real-world environments. Typical applications include scheduling, production planning, logistics, network design, communication and routing in networks, and health care. Her research has won several awards, including the Heinz Maier-Leibnitz Prize in 2013. 

https://www.uni-bremen.de/en/cslog/nmegow 

Evaluation of Health Interventions and Predictions 

(Werner Brannath)

The working group "Applied Statistics and Biometry" at the Faculty Mathematics/Computer Science at the University of Bremen is a group of mathematicians and statisticians working on the application, investigation and development of statistical methods for the life sciences. We also run (together with the BIPS) the international master programme "Medical Biometry/Biostatistics" and teach mathematics master students. We are involved in the Competence Centre for Clinical Trials Bremen (KKSB), where we support physicians and health scientists in the planning and statistical analysis of studies evaluating health and care interventions. Our current methodological research includes adaptive and group sequential designs in precision medicine. Furthermore, we develop multiple testing and simultaneous inference methods for clinical trials, for online multiple testing as well as for a more efficient evaluation of classification and machine learning algorithms. In my presentation, I will illustrate our research activities with examples. This will include an ongoing study on the implementation and evaluation of a new medical care structure for the treatment and management of chronic wounds in the state of Bremen. I will also present examples of some of our recently developed statistical methods for clinical trials, online multiple testing and the evaluation of classification algorithms. I will also give an outlook on future research plans and projects.

Bio:

Werner Brannath has studied mathematics and physics at the Universities of Karlsruhe and Vienna. He was Research Assistant at the Department of Statistics at the University of Vienne and Assistant Professor at the Institute of Medical Statistics of the Medical Faculty/Medical University of Vienna (MUW). After his habilitation he became Associated Professor at the MUW. In 2001 he received an Erwin Schrödinger scholarship for a one year research stay at the Department of Statistics at Stanford University. Since August 2010 he is Professor of Applied Statistics and Biometry at the University of Bremen (Faculty 3) and head of the Biometry Section of the Competence Center for Clinical Trials Bremen (KKSB). He is also a member of the KKSB’s collegial leadership. In March 2017 he was elected as president of the German Region of the International Biometric Society (IBS-DR).

Besides his activities as responsible biostatistician (responsible for the biostatistics and data management) in a number of medical and health science research projects, he is developing and investigating statistical methods e.g. for sequential und adaptive designs and experiments in the engineering sciences as well as for simultaneous statistical inference. In the last years he also did research on variable selection procedures and methods for the evaluation of machine learning predictors. In a current research project he works on methods for statistical inference after model- und variable selection.  

https://www.uni-bremen.de/en/kksb/biometrie/team/members/prof-dr-werner-brannath