How to calculate sensitivity and specificity from Deep Network trained data?

I have trained a data with my network arhitecture and i got an accuracy of 75.6%. I want to calculate specificity and sensitivity using trained Info. My final validation loss is 0.5328 and final validation accuracy is 75.6.

Answer 1

[Jayanti]

Hi Theodora,

To calculate sensitivity and specificity for a trained neural network you can follow the below steps :-

1. Ensure you have a test dataset with true labels and predictions from your neural network.
2. Use the trained network to make predictions.
3. Then identify the number of cases for True positive, True negative, False positive and False negative.
4. Use below formulas to calculate sensitivity and specificity

sensitivity = TP / (TP + FN)

specificity = TN / (TN + FP)

y_true = [1, 0, 1, 1, 0, 1, 0, 0, 1, 0]; 
y_pred = [1, 0, 1, 0, 0, 1, 1, 0, 1, 0]; 
confMat = confusionmat(y_true, y_pred); 
TP = confMat(2, 2); 
TN = confMat(1, 1); 
FP = confMat(1, 2); 
FN = confMat(2, 1); 
sensitivity = TP / (TP + FN); 
specificity = TN / (TN + FP); 

You may refer to the below MathWorks documentation to know more on “confusionmat” function:

https://in.mathworks.com/help/stats/confusionmat.html