N-of-1 Trial Methodology
Simulation-based evaluation of N-of-1 versus parallel-group designs for biomarker-validated treatments
2026-06-07 18:16 PDT
Overview
N-of-1 trials – crossover designs in which a single subject serves as their own control across multiple treatment periods – offer substantial power advantages when treatment effects are heterogeneous across individuals. For chronic diseases such as Alzheimer’s and PTSD, where individual response patterns are highly variable and biomarker-validated patient stratification is increasingly feasible, N-of-1 designs deserve systematic methodological evaluation against conventional parallel-group randomized controlled trials.
This program develops a coordinated series of simulation studies comparing N-of-1 and parallel-group architectures across the design dimensions most consequential for Alzheimer’s disease and PTSD trials: carryover effects, serial correlation, dropout patterns, biomarker- treatment interaction, and the statistical procedures applied at the analysis stage.
Current work
The core of this program is a nine-paper Statistics in Medicine manuscript series, each sharing a common R simulation package (pmsimstats, v0.3) and a unified staged-render toolchain:
- Paper 1. Two architectures for simulating biomarker-treatment interactions: implications for statistical power under carryover.
- Paper 2. Robustness of carryover-mitigation analysis strategies: a factorial simulation study.
- Paper 3. Latent-class and mixture-model formulations for biomarker-treatment interaction in N-of-1 trials.
- Paper 4. Comparative statistical power of N-of-1 trials versus RCTs: a simulation study of prazosin for PTSD nightmares.
- Paper 5. Power sensitivity of a hybrid open-label-plus-blinded- discontinuation N-of-1 design under carryover and serial correlation.
- Paper 6. Three-component decomposition of treatment response in aggregated N-of-1 trials: pharmacological, expectation-driven, and natural-history components.
- Paper 7. Critical evaluation of the modified Gompertz response function as a parametric template for treatment-response trajectories.
- Paper 8. Test-procedure and trial-design choices for the biomarker-treatment interaction: classical RM-ANOVA, mixed-effects, and the cycle-by-period protocol grid.
- Paper 9. Informative dropout and trial-design choice in aggregated N-of-1 biomarker-validation trials.
All nine manuscripts have rendered draft PDFs; the series targets Statistics in Medicine.
Methods
Monte Carlo simulation under the ADEMP framework; linear mixed-effects models (MMRM and RM-ANOVA); latent-class and mixture models; Gompertz response functions; empirically derived variance-covariance matrices from ADCS trials; carryover-sensitivity factorial designs.
Software
- nof1power – Power analysis and simulation for N-of-1 trials, implementing the core simulator described in the paper series.
Publications
This is a primarily methodological program; the nine-paper series is the principal output and is in preparation for submission. Earlier work on N-of-1 design and power is accessible through the full publications list by filtering on the keyword simulation.