Abstract
<jats:p><p>Motor evoked potential (MEP) latencies generated in transcranial magnetic stimulation (TMS) experiments are an important and fundamental physiological parameter for studying the human motor system. Manual annotation of MEPs can be performed, but this is time-consuming and subject to human error, which can lead to problems with automated methods for extracting MEP features. The purpose of this study is to compare existing methods for automated MEP latency determination in terms of their advantages and disadvantages and propose a new method that can improve results. The hypothesis is that we can propose a new automated annotation model that can improve results, or that this model will fail to automatically extract MEP features. Each existing method will be examined in terms of its advantages and disadvantages to identify targeted issues that will be addressed in the proposed model. The results show that all existing methods share a common problem: they are not reproducible to other datasets. This means that each model was focused on extracting MEP features under specific conditions and dataset formats. This study identified relevant challenges that we must consider when developing a new model for automatically annotating MEP features. Therefore, it is recommended to pay attention to every detail to achieve successful results.</p></jats:p>