Skip to main content

Table 2 Common metrics in AI model evaluation

From: Application of machine learning in ophthalmic imaging modalities

Evaluation metrics

Definitions

Accuracy

The proportion of both positives and negatives that are correctly identified; the higher the accuracy, the better the classifier

Sensitivity/Recall

The proportion of positives that are correctly identified

Specificity

The proportion of negatives that are correctly identified

Precision

The proportion of positives that are correctly identified among all positive identified samples

Kappa value

To show the actual agreement between two sets of observations

Dice coefficient/F1 score

Harmonic average of the precision and recall, where an F1 score reaches its best value at 1 and worst at 0