This study determined the efficacy of an explainable machine learning method in analyzing electrocardiogram (ECG) data for left ventricle (LV) scar detection in patients with hypertrophic cardiomyopathy (HCM). The machine learning model, XplainScar, was trained on cardiac MRIs and 12-lead ECGs from 748 patients with HCM combined from two datasets. They integrated an explanation framework that allows users to determine how the model weighed each variable in its final decision of LV scar or no scar. The model was internally validated with just ECG data from 500 patients. It was found that the model could effectively detect LV scar with a sensitivity of 0.95 and an F1 score of 0.92 in its training dataset, and a sensitivity of 0.91 and an F1 score of 0.89 in its validation dataset. It was able to come to its reliable conclusion in less than a minute with only one 10-second 12-lead ECG per patient. The model was most accurate when there was a high LV scar burden. A benefit of the model is its explanation framework, which allowed the authors to find that even when values were not significantly different between the Scar and NoScar groups, they were still used by the model to make its decision. This shows that the model was able to use both common HCM markers, such as deep T wave inversions, while also finding new relationships in less common variables, such as intrinsicoid deflection, when determining LV scar.