A0385
Title: Comparing F1-scores of more than two binary medical tests
Authors: Kanae Takahashi - Hyogo Medical University (Japan) [presenting]
Abstract: In the medical field, binary classification problems are common, and accuracy, sensitivity, specificity, and negative and positive predictive value are often used as indicators of the performance of binary medical tests. Additionally, the F1-score is often used in the field of information retrieval, and it is gaining popularity in the medical field. This score is defined as the harmonic mean of recall (sensitivity) and precision (positive predictive value). A statistical test procedure to compare two F1-scores was recently proposed. However, it is often the case that more than two F1-scores are reported and considered at the same time, and it may be desirable to compare them simultaneously. Therefore, using the multivariate central limit theorem and the delta-method, we developed a test procedure for comparing F1-scores of more than two binary medical tests simultaneously.
