# Precision, accuracy, and reliability of a threshold hunting method for transcranial magnetic stimulation

## TL;DR

A new method for estimating motor threshold in transcranial magnetic stimulation is precise, accurate, and reliable within 18 trials.

## Contribution

The study demonstrates that the PEST-MLE method achieves reliable RMT estimation with high precision and accuracy in a small number of trials.

## Key findings

- The PEST-MLE method achieved good within-day reliability with ICC values >0.80.
- RMT estimation converged in precision and accuracy by the 18th trial.
- Reproducibility coefficient and standard error were around 10% and 5%, respectively.

## Abstract

•Adaptive method estimated motor threshold precisely and accurately in 18 trials.•Within-day reliability was good: intraclass correlation coefficients were >0.80.•Reproducibility coefficient was around 10 % with the standard error around 5 %

Adaptive method estimated motor threshold precisely and accurately in 18 trials.

Within-day reliability was good: intraclass correlation coefficients were >0.80.

Reproducibility coefficient was around 10 % with the standard error around 5 %

To investigate precision, accuracy, and reliability of a threshold hunting method to estimate resting motor threshold (RMT) using parameter estimation by sequential testing and maximum likelihood estimation (PEST-MLE).

In 53 healthy participants, single pulse transcranial magnetic stimulation was delivered to estimate RMT with cutoff values of 0.05 mV and 0.2 mV. RMT was inferred from 30-trial PEST-MLE algorithm for at maximum two days with two estimation sessions per day, comprising Sessions 1 to 4. Precision and accuracy were assessed within each session. Repeatability was assessed using intraclass correlation coefficient (ICC), reproducibility coefficient (RC), and standard error of measurement (SEM).

For both 0.05 mV and 0.2 mV, 18 trials were needed to have good accuracy. ICC greater than 0.8 was achieved for within-day comparison but the ICC of between-day comparison was lower. RC and SEM were around 10 % and 5 %, respectively.

At 18th trial the estimations were converged in terms of precision and accuracy, and good reliability was achieved at that stage.

RMT estimation with the PEST-MLE is a rapid and reliable approach that can be implemented for future clinical and scientific studies.