Aim and content

This is a generic course. This means that the course is reserved for PhD students at the Graduate School of Health and Medical Sciences at UCPH.

Anyone can apply for the course, but if you are not a PhD student at the Graduate School, you will be placed on the waiting list until enrollment deadline. After the enrolment deadline, available seats will be allocated to the waiting list.

The course is free of charge for PhD students at Danish universities (except Copenhagen Business School), and for PhD students at NorDoc member faculties. All other participants must pay the course fee

Learning objectives
A student who has met the objectives of the course will be able to:

1. Explain the fundamental principles of statistical inference using targeted minimum loss-based estimation (TMLE) and its application as a general framework for estimation of causal effects.
2. Implement TMLE using R software to estimate average treatment effects and time-varying treatment effects based on simulated data, and assess the accuracy and efficiency of the estimators.
3. Compare the assumptions and performance of TMLE to related causal inference tools such as inverse probability weighting and standardization, and discuss the strengths and limitations of each approach.
4. Evaluate the suitability of super learning and its application in TMLE, and implement the algorithm to improve estimation accuracy.
5. Discuss and evaluate the challenges and opportunities in time-varying settings in causal inference, including time-varying treatments and time-dependent confounding, and how TMLE can be used to address these challenges.

Content
Targeted minimum loss-based estimation (TMLE) is a general framework for estimation of causal effects that combines semiparametric efficiency theory and machine learning in a two-step procedure. The main focus of the course is to understand the overall concept, the theory, and the application of TMLE. Topics covered include:
• The roadmap of targeted learning.
• Basics of causal inference, including counterfactual notation, hypothetical interventions, the g-formula, and the average treatment effect (ATE).
• Causal effect estimation in nonparametric models: target parameters, nuisance parameters, efficient influence functions, asymptotic linearity, and statistical inference based on the efficient influence function.
• TMLE as a two-step procedure involving initial estimation followed by a targeting step.
• Super learning: combining multiple machine learning algorithms via loss-based cross-validation.
• Extensions to more complex data settings: survival outcome, time-dependent confounding, dynamic treatment regimes.
• Basic usage of existing software in R.

Course material: A combination of research papers, textbook, and lecture notes.

Participants
The course is relevant for Ph.D.-students with sufficient background in mathematics and statistics. To participate in the practicals, the participants should have knowledge of the statistical software R.
Max. 25 participants.

Relevance to graduate programmes
The course is relevant to PhD students from the following graduate programmes at the Graduate School of Health and Medical Sciences, UCPH:

All graduate programmes

Language
English

Form
Four days with lectures and practicals.

Course director
Helene Charlotte Wiese Rytgaard, Assistant Professor, Section of Biostatistics, Department of Public Health.

Teachers
Helene Charlotte Wiese Rytgaard, Thomas Alexander Gerds, Anders Munch

Dates
10, 11, 12, 13 June 2025; all days 8-15

Course location
CSS

Registration
Please register before 9 May 2025

Seats to PhD students from other Danish universities will be allocated on a first-come, first-served basis and according to the applicable rules.
Applications from other participants will be considered after the last day of enrolment

Note: All applicants are asked to submit invoice details in case of no-show, late cancellation or obligation to pay the course fee (typically non-PhD students). If you are a PhD student, your participation in the course must be in agreement with your principal supervisor.

Disclaimer:
DDSA has explicit permission from Arcanic and the owners of the https://phdcourses.dk/ website to display the courses on ddsa.dk.