Name: How many patients do you need? A trial-design simulation for anti-seizure treatment in acute brain injury
Published: July 11, 2026
License: https://github.com/bdsp-core/bdsp-license-and-dua

Software Credentialed Access

Harsh Parikh , Haoqi Sun , Rajesh Amerineni , Eric Rosenthal , Alexander Volfovsky , Cynthia Rudin , M. Brandon Westover , Sahar F. Zafar

Published: July 11, 2026. Version: 1.0.0

When using this resource, please cite: (show more options)
Parikh, H., Sun, H., Amerineni, R., Rosenthal, E., Volfovsky, A., Rudin, C., Westover, M. B., & Zafar, S. F. (2026). How many patients do you need? A trial-design simulation for anti-seizure treatment in acute brain injury (version 1.0.0). Brain Data Science Platform. https://doi.org/10.60508/chjy-bf16.

MLA	Parikh, Harsh, et al. "How many patients do you need? A trial-design simulation for anti-seizure treatment in acute brain injury" (version 1.0.0). Brain Data Science Platform (2026), https://doi.org/10.60508/chjy-bf16.
APA	Parikh, H., Sun, H., Amerineni, R., Rosenthal, E., Volfovsky, A., Rudin, C., Westover, M. B., & Zafar, S. F. (2026). How many patients do you need? A trial-design simulation for anti-seizure treatment in acute brain injury (version 1.0.0). Brain Data Science Platform. https://doi.org/10.60508/chjy-bf16.
Chicago	Parikh, Harsh, Sun, Haoqi, Amerineni, Rajesh, Rosenthal, Eric, Volfovsky, Alexander, Rudin, Cynthia, Westover, M. Brandon, and Sahar F. Zafar. "How many patients do you need? A trial-design simulation for anti-seizure treatment in acute brain injury" (version 1.0.0). Brain Data Science Platform (2026). https://doi.org/10.60508/chjy-bf16.
Harvard	Parikh, H., Sun, H., Amerineni, R., Rosenthal, E., Volfovsky, A., Rudin, C., Westover, M. B., and Zafar, S. F. (2026) 'How many patients do you need? A trial-design simulation for anti-seizure treatment in acute brain injury' (version 1.0.0), Brain Data Science Platform. Available at: https://doi.org/10.60508/chjy-bf16.
Vancouver	Parikh H, Sun H, Amerineni R, Rosenthal E, Volfovsky A, Rudin C, Westover M B, Zafar S F. How many patients do you need? A trial-design simulation for anti-seizure treatment in acute brain injury (version 1.0.0). Brain Data Science Platform. 2026. Available from: https://doi.org/10.60508/chjy-bf16.

Additionally, please cite the original publication:

Parikh H, Sun H, Amerineni R, Rosenthal ES, Volfovsky A, Rudin C, Westover MB, Zafar SF. How many patients do you need? Investigating trial designs for anti-seizure treatment in acute brain injury patients. Ann Clin Transl Neurol. 2024;11(7):1681-1690. PMID: 38867375.

Abstract

Objective. Designing randomized trials of anti-seizure treatment in critically ill patients is hard because seizure burden and drug response are highly heterogeneous. We built a simulation framework to estimate how many patients a trial would need under various designs.

Methods. Using continuous-EEG and medication data from a subarachnoid-hemorrhage cohort, we fit patient-level log-normal models of seizure/spike dynamics and drug response, then simulated randomized trials to compute required sample sizes and detectable effect sizes for outcomes such as complete seizure cessation, across treatment-delay and drug-response scenarios.

Results. The framework quantifies the sample size and effect size needed under each design, showing how trial feasibility depends on the outcome definition and response assumptions.

This project releases the de-identified patient-level parameters and time series and the simulation code.

Background

Randomized trials of anti-seizure treatment in acute brain injury are difficult to power because seizure burden and drug response vary widely across patients. A simulation framework grounded in real patient-level EEG and drug data can estimate the sample sizes needed under different trial designs.

Software Description

MATLAB pipeline (Step_1 parameter fitting, Step_2 sample-size/effect-size simulation, Drug_Outcomes) + Python curve-fitting, and de-identified per-patient data keyed by sidNNNN: fitted parameters (Parameters 48 patients.xlsx), spike/CNN-label/normalized-drug time series, and the committed simulation results. No MRNs, names, or dates.

Technical Implementation

Patient-level log-normal models of seizure/spike dynamics and drug response were fit from continuous-EEG and medication-administration data in a subarachnoid-hemorrhage cohort, then used to simulate randomized trials and compute required sample sizes and detectable effect sizes across outcome definitions and response/delay scenarios.

Installation and Requirements

MATLAB R2018+ (Step_1/Step_2); Python 3 with numpy/scipy for Curve_fit_Python.py. Run Step_2_Calculate_Sample_Effect_Size from the repo root. See REPRODUCE.md and DATA_SOURCE.md.

Usage Notes

Step_2_Calculate_Sample_Effect_Size reproduces the sample-size/effect-size results from the committed 48-patient parameters (a Monte-Carlo simulation); Sim_Results_Store holds the paper's committed outputs; figure code is in Code_To_Generate_Figures_Paper/.

Release Notes

First public release: simulation code + de-identified patient-level parameters/time series.

Ethics

De-identified data under IRB approval; raw medication records and MRN/name linkage excluded; only sidNNNN-keyed derived data released.

Acknowledgements

Data from the MGH SAH SAGE continuous-EEG database.

Conflicts of Interest

See the associated publication (Ann Clin Transl Neurol 2024;11:1681-1690).

References

Parikh H, Sun H, Amerineni R, Rosenthal ES, Volfovsky A, Rudin C, Westover MB, Zafar SF. How many patients do you need? Investigating trial designs for anti-seizure treatment in acute brain injury patients. Ann Clin Transl Neurol. 2024;11(7):1681-1690. PMID: 38867375.