Model-Twin Randomization (MoTR) for Estimating the Recurring Individual Treatment Effect
Last updated: Nov 18, 2025
How it started (2006):
Four years. The time between the summer Olympics, between leap years. The time from freshman orientation to graduation day. The time it took for our manuscript to go from methodological “garage startup” to bona fide publication.
I loved writing this paper—so much! I am incredibly proud of what we accomplished together. ❤️
Getting published in Statistics in Medicine (Stat Med) is a true delight. Stat Med is one of the most competitive, rigorous, and well-regarded scientific journals in applied biostatistics and medical statistics. A first-authored Stat Med publication shows that I am a serious methodologist—that I have finally arrived. 😎🥰
Manuscript: Daza, Eric J., Igor Matias, and Logan Schneider. “Model-Twin Randomization (MoTR) for Estimating the Recurring Individual Treatment Effect.” Statistics in Medicine 44.25-27 (2025): e70290. onlinelibrary.wiley.com/doi/10.1002/sim.70290 | tinyurl.com/dazaetal2025
Manuscript (full text): onlinelibrary.wiley.com/share/author/KXXU7QDSACC6YVK7XKSM?target=10.1002/sim.70290 | tinyurl.com/dazaetal2025fulltext | tinyurl.com/dazamotr
Lay summary (2-minute video): tinyurl.com/dazaetal2025video
ABSTRACT: Temporally dense single-person “small data” have become widely available thanks to mobile apps and wearable sensors. Many caregivers and self-trackers want to use these data to help a specific person change their behavior to achieve desired health outcomes. Ideally, this involves discerning possible causes from correlations using that person’s own observational time series data. In this paper, we estimate within-individual average treatment effects of physical activity on sleep duration. We introduce the model twin randomization (MoTR; “motor”) method for analyzing an individual’s intensive longitudinal data. Formally, MoTR is an application of the g-formula (i.e., standardization, back-door adjustment) under serial interference. It estimates stable recurring individual treatment effects, as is done in n-of-1 trials and single case experimental designs. We compare our approach to standard methods (with possible confounding) to show how to use causal inference to make better personalized recommendations for health behavior change, and analyze up to almost eight years of the authors’ own Fitbit steps and sleep data.
- Story behind the manuscript: Once Upon a Time Series (biostatistics.ca/once-upon-a-time-series)
- MoTR Python code (by Igor Matias): gitlab.unige.ch/qol/MoTR-python (R version forthcoming)
- MoTR digital health application: Matias, Igor, Eric J. Daza, and Katarzyna Wac. “What possibly affects nighttime heart rate? Conclusions from N-of-1 observational data.” Digital Health 8 (2022): 20552076221120725. journals.sagepub.com/doi/10.1177/20552076221120725
My coauthors and I had built a well-tempered instrument—a precision timepiece that was as technically delightful to design, contemplate, and iterate, as it (hopefully!) is to practically implement. It’s ornate yet simple, a didactic gift to curious scientists and methodologists.
But this scientific article is also part of a growing call to fundamentally change health and medicine research, diagnosis, and treatment: From relying on misnamed “precision” and “personalization” approaches that in fact characterize small groups (not individuals), to investigating and shaping truly individualized patterns unique to each person. It is a call to think beyond statistics—to also think in terms of esametrics, “the application of statistics to a single person, individual, or unit” (Daza et al, 2025) that includes n-of-1 trials and single-case designs.
You might be interested in our manuscript if you like:
- causal inference, real-world evidence, heterogeneous treatment effects, conditional average treatment effects, individualized treatment effects
- n-of-1 trials, single-case designs, within-subject designs, switchback experiments
- time series, functional data analysis, microsimulation, control theory
- precision medicine, personalized health, patient-centricity, patient reported outcomes
- digital health, digital endpoints, digital twins
- chronic health conditions, chronic disease prevention and management
- sports and athletic performance, sports medicine
There are so many wonderful people to thank! These include my coauthors Igor Matias and Logan Schneider; professors Linda Valeri, Mike Baiocchi, Jared Huling, and Katarzyna Wac for their guidance; our two excellent paper referees at Statistics in Medicine; Luca Foschini and many others at Evidation; my fabulous Stats-of-1 co-editors Clair Robbins and Julio Vega; and media colleagues who helped spread the word about Stats-of-1 or MoTR. The latter include content creators Glen Colopy, Justin Belair, Alex Molak, Aline Holzwarth (Forbes Magazine) and Kasandra Brawaw (Fortune Magazine). And, of course, I am deeply grateful to all my family and friends—y’all are the best!
I dedicate this paper to Filipinos and Filipino-Americans. To all who are underrepresented or unacknowledged in science, technology, engineering, and math (STEM), and in academia: Kaya natin ’to! To you, the reader: Know yourself, help others, and find meaning in all things.