ARC@ORU: Vad är framtiden för effektuppskattning med big data?
06 november 2025 16:00 – 17:00 Visual Lab i ARC-huset, eller digitalt via Zoom
Välkommen till ett seminarium inom ARC:s tillämpningsområde Life Science and Health. Forskaren Johannes Stork går igenom grunderna för effektuppskattning, big data och dess utmaningar, och hur metoder inom deep learning kan användas.
Seminariet hålls på engelska / The Seminar is held in English
What is the future of effect estimation with big data?
This is a seminar within the ARC horizontal Life Science and Health.
- Speaker: Johannes Stork, Associate Professor of Computer Science,
Head of Adaptive and Interpretable Learning Systems Lab, Örebro University - Hosts: Marcus Krantz, Biomedicine Pedro Zuidberg dos Martires, Computer Science
- Time: 6 November, kl 16.00-17.00
- Venue: Visual Lab (Please register further down on the webpage)
- Remote participation: Use this Zoom-link
About the seminar
Big data is often collected in an uncontrolled process and can therefore contain selection bias. To make use of it, we try to remove this bias. However, with large amounts of data and many variables, classical approaches struggle. This is oftentimes due to problems of identifying causal factors and relevant variables. Deep representation learning approaches tries to answer this call.
In this talk, we will look at the basics of effect estimation, move to the big data setting and its challenges, and see how deep learning methods can be used in this setting. Finally, we will ask ourselves, but is this the right way to go?
Proper treatment effect estimation is a crucial challenge in the life sciences and health, as it underpins the evaluation of clinical interventions. Reliable estimates enable clinicians and policymakers to make evidence-based decisions about which treatments improve outcomes and avoid those that are ineffective or harmful.
This is relevant for the AI, Robotics & Cybersecurity Center as it connects AI and statistical learning with healthcare by advancing causal inference, which is a key step toward trustworthy decision support, reliable clinical analytics, and safe health applications.