Twenty years of Targeted Learning

Nima Hejazi

nhejazi@hsph.harvard.edu

Harvard Biostatistics

May 13, 2026

Session overview

We aim to review perspectives on theory, methods, and applications of Targeted Learning, a statistical methodology framework that blends

• causal inference, to translate scientific questions into estimands;
• machine learning, for flexible estimation of parameter components; and
• semi-parametric theory, for uncertainty quantification.

Targeted Learning (TMLE) was first pioneered in 2006, with new theory and methods continuing to appear at a rapid pace and frequent applications in health sciences, ranging from vaccine efficacy trials to precision psychiatry to population-scale genomics.

Targeted Learning—An abbreviated timeline

• Initial framework: Targeted maximum likelihood learning (van der Laan and Rubin, 2006)
• Ensemble machine learning: Super Learner (van der Laan et al., 2007)
• Extensions to settings with time-varying confounding: L-TMLE (Stitelman et al., 2011; van der Laan, 2010; van der Laan and Gruber, 2011)
• Theoretical arguments for use of cross-fitting: CV-TMLE (Zheng and van der Laan, 2011)
• First book reviewing methodological advances: Targeted Learning: Causal Inference for Observational and Experimental Data (van der Laan and Rose, 2011)
• Advances in semi-parametric theory: Doubly robust inference (Benkeser et al., 2017; Dı́az and van der Laan, 2017; van der Laan, 2014)
• And back to non-parametric maximum likelihood: Highly adaptive lasso (HAL) (Benkeser and van der Laan, 2016; van der Laan, 2017)
• Second book reviewing methodological advances: Targeted Learning in Data Science: Causal Inference for Complex Longitudinal Studies (van der Laan and Rose, 2018)
• Data-adaptively learning a working statistical model: A-TMLE (van der Laan et al., 2023)

Speakers and talks

1. Mark van der Laan (University of California, Berkeley)
   My journey from NP-MLE to the current state of Targeted Learning
2. Alex Luedtke (Harvard Medical School)
   DoubleGen: Debiased generative modeling of counterfactuals
3. Mireille Schnitzer (Universite de Montreal)
   Longitudinal outcome-HAL with doubly robust inference LTMLE
4. Sjoerd Beentjes & Ava Khamseh (University of Edinburgh)
   Targeted Learning for genomics in large-scale cohorts
5. Brief panel discussion with all speakers

References

Benkeser, D. and van der Laan, M. J. (2016) The highly adaptive lasso estimator. In: 2016 IEEE international conference on data science and advanced analytics (DSAA), 2016, pp. 689–696. IEEE.

Benkeser, D., Carone, M., van der Laan, M. J., et al. (2017) Doubly robust nonparametric inference on the average treatment effect. Biometrika, 104, 863–880. Oxford University Press. DOI: 10.1093/biomet/asx053.

Dı́az, I. and van der Laan, M. J. (2017) Doubly robust inference for targeted minimum loss-based estimation in randomized trials with missing outcome data. Statistics in Medicine, 36, 3807–3819. Wiley Online Library. DOI: 10.1002/sim.7389.

Stitelman, O. M., De Gruttola, V. and van der Laan, M. J. (2011) A general implementation of TMLE for longitudinal data applied to causal inference in survival analysis. 281, April. University of California, Berkeley. Available at: https://biostats.bepress.com/ucbbiostat/paper281.

van der Laan, L., Carone, M., Luedtke, A., et al. (2023) Adaptive debiased machine learning using data-driven model selection techniques. arXiv preprint arXiv:2307.12544.

van der Laan, M. J. (2010) Targeted maximum likelihood based causal inference: Part i. International Journal of Biostatistics, 6, 2. De Gruyter. DOI: 10.2202/1557-4679.1211.

van der Laan, M. J. (2014) Targeted estimation of nuisance parameters to obtain valid statistical inference. The International Journal of Biostatistics, 10, 29–57. De Gruyter. DOI: 10.1515/ijb-2012-0038.

van der Laan, M. J. (2017) A generally efficient targeted minimum loss based estimator based on the highly adaptive lasso. International Journal of Biostatistics, 13. De Gruyter.

van der Laan, M. J. and Gruber, S. (2011) Targeted minimum loss based estimation of an intervention specific mean outcome. 290, August. University of California, Berkeley. Available at: https://biostats.bepress.com/ucbbiostat/paper290.

van der Laan, M. J. and Rose, S. (2011) Targeted Learning: Causal Inference for Observational and Experimental Data. Springer. DOI: 10.1007/978-1-4419-9782-1.

van der Laan, M. J. and Rose, S. (2018) Targeted Learning in Data Science: Causal Inference for Complex Longitudinal Studies. Springer. DOI: 10.1007/978-3-319-65304-4.

van der Laan, M. J. and Rubin, D. (2006) Targeted maximum likelihood learning. The International Journal of Biostatistics, 2. De Gruyter. DOI: 10.2202/1557-4679.1043.

van der Laan, M. J., Polley, E. C. and Hubbard, A. E. (2007) Super learner. Statistical Applications in Genetics and Molecular Biology, 6. De Gruyter. DOI: 10.2202/1544-6115.1309.

Zheng, W. and van der Laan, M. J. (2011) Cross-validated targeted minimum-loss-based estimation. In Targeted Learning: Causal Inference for Observational and Experimental Data (eds. M. J. van der Laan and S. Rose), pp. 459–474. Springer. DOI: 10.1007/978-1-4419-9782-1_27.