AMDS@Pacmed #3: Combining causal inference and machine learning in practice

Organised by Pacmed (Pacmed.ai) in collaboration with Freedomlab (freedomlab.nl).

How can we combine the fields of causal inference and machine learning? Building the best predictive model in healthcare can be totally different from building the best explanatory model. During this meetup we’d like to discuss the opportunities and challenges in combining the fields of causal inference and machine learning, deepdive into the difficulties that come with the approach(es) using observational data but also discuss methods that allow to combine those two fields in the best way possible. Join us for an applied discussion with Gertjan (data scientist) & Misja (Chief economist) from the Nederlandse Zorgautoriteit and Christina & Hans (both data scientists) from Pacmed.

PROGRAM:

18.00 - 18.30 Doors open & Welcome
18.30 - 19.15 Gertjan Verhoeven/ Misja Mikkers- The effect of Tilburg gold on your health
19.15 - 19.30 Break
19.30 - 20.15 Christina Stamm/ Hans de Ferrante- Estimating causal effects of adjuvant breast cancer therapy on overall survival
20.15 - 21.00 Drinks & Networking

There will be some (small) snacks and drinks before and after the event.

Specific details on speakers bio and content to be announced.

ABSTRACTS:

The effect of Tilburg gold on your health
In this presentation we introduce causal graphical models as a tool for inference. We demonstrate how to build and analyze causal graphical models with the tool Dagitty. We will show the consequences of switching from prediction to causal inference for machine learning analysis.

Estimating causal effects of adjuvant breast cancer therapy on overall survival
Details TBA

ABOUT THE SPEAKERS:

Gertjan Verhoeven and Misja Mikkers both work at the Dutch Healthcare Authority (NZa).

Dutch healthcare is a highly complex system of interacting healthcare providers, health insurers, consumers and a diverse circle of stakeholders and authorities. They should be able to rely on good and affordable healthcare if and when they need it. The NZa makes and enforces rules and monitors healthcare. We make extensive use of data analysis and models. The NZa uses these for monitoring developments in the market, conduct research and create policies to prevent or solve social problems.

Gertjan Verhoeven is a research scientist, working on health policy and statistical methods He has a background in experimental physics, with a PhD in Biophysics. Misja Mikkers is chief economist of the NZa. He combines this role with an appointment as endowed professor at Tilburg University.

Christian Stamm and Hans de Ferrante both work at Pacmed- where they apply causal inference and machine learning in practice to develop decision support tools.